Introduction

host—JORDAN CHENEVIER-TRUCHET:
Selma, tu es la première personne à venir deux fois dans Conquête, donc c'est vraiment cool. La dernière fois qu'on s'était parlé, tu n'étais pas chez Agorapulse. Tout à fait. Et tu n'étais pas encore CRO. Tout à fait. Donc ça m'intéresse que tu puisses commencer par nous expliquer ce shift.
CRO, qu'est-ce que ça veut dire dans ton job ? Et un autre sujet qui peut être intéressant : généralement, les personnes promues au poste de CRO viennent plutôt du monde du sales. Toi, dans ton cas, c'est plutôt le marketing, donc ça m'intéresse de comprendre pourquoi.
Déjà, pourquoi tu penses que c'est pertinent ? C'est quoi l'avantage que tu as par rapport à quelqu'un qui viendrait du sales ? Voilà un peu ce qui va m'intéresser avec ce nouveau rôle.

guest—SELMA CHAUVIN—CRO Agorapulse:
OK, donc oui, ça fait maintenant 18 mois que je suis CRO chez Agorapulse. Dans le scope, qu'est-ce que j'ai ? J'ai le marketing traditionnel, j'ai aussi le RevOps, j'ai les équipes commerciales, j'ai le key account management et j'ai le new business ultra classique. La chose qui ne me plaît pas, parfois, dans les équipes revenue, c'est que la partie CS est à part, avec le support.

host—JORDAN CHENEVIER-TRUCHET:
Pourquoi tu ne les as pas ?

guest—SELMA CHAUVIN—CRO Agorapulse:
Parce que je pense déjà que je n'ai aucune compétence dedans, ce qui est déjà une bonne première étape. Deuxième chose, mon travail est vraiment focus sur le revenu, donc à partir de mon scope, sur la partie suivi du client au quotidien, je suis beaucoup moins pertinente.

host—JORDAN CHENEVIER-TRUCHET:
OK, d'accord. Et donc t'étais CMO ?

guest—SELMA CHAUVIN—CRO Agorapulse:
Oui. Donc pourquoi je suis passée à CRO ? Alors tout ça vient d'un déjeuner, parce qu'évidemment ce n'est pas un parcours hyper naturel. Comme tu le disais, on voit plutôt un profil sales à qui on propose de prendre le poste de CRO, parce qu'on voit le marketing comme optionnel et que ce qui va compter avant tout, c'est la capacité à closer.
Sauf qu'en fait, la majeure partie des boîtes, quand tu leur parles, n'ont pas un problème de closing, elles ont un problème de pipe. Et avec ce prisme-là, si tu prends un profil sales qui devient CRO, qu'est-ce que tu fais ? Tu mets juste plus de pression et tu matraques encore plus ton marketing sur du résultat à court terme.
Alors que si tu prends vraiment l'autre approche, qui dit "j'ai un problème de pipe, j'en ai conscience, je vais prendre quelqu'un qui sait gérer un problème de pipe", eh bien tu vas embaucher côté marketing. Ça a commencé comme ça. Avec Aymeric, qui était le CEO d'Agorapulse à l'époque, on parlait tous les trimestres de comment ça se passait côté marketing, quelle était notre poigne sur le pipe, comment on pouvait résoudre ça.
Et de fil en aiguille, on a discuté, et ça a fini par rejoindre ce scope, et essayer de régler le problème, pour moi.

host—JORDAN CHENEVIER-TRUCHET:
OK, mais du coup, la manière dont tu décris les choses, je me dis que finalement tu aurais pu prendre un poste de CMO : tu as vu que le problème était un problème de pipe, et pourtant tu as pris aussi un scope qui est très sales.

guest—SELMA CHAUVIN—CRO Agorapulse:
Oui, parce que c'est là la problématique. Quand tu embauches un CMO alors que t'as un problème de pipe... je te pose la question à toi.

host—JORDAN CHENEVIER-TRUCHET:
C'est une vraie question. Qu'est-ce qui se passe quand t'en as un ? Si un CMO a un problème de pipe, eh bien c'est focus lead, et la partie conversion passe souvent un peu à la trappe, c'est ça ?

guest—SELMA CHAUVIN—CRO Agorapulse:
Donc la problématique, c'est que tu ne vas pas régler ton problème de lead : à la rigueur, tu peux avoir plein de leads qui attendent sur l'étagère que quelqu'un s'en occupe. Il te faut vraiment quelqu'un qui ait la main complète, capable d'aller taper partout en disant : hey, t'as fait quoi des leads qu'on t'a donnés ?

host—JORDAN CHENEVIER-TRUCHET:
Oui, et il y a aussi la réflexion sur la valeur des leads qui entre en compte.

guest—SELMA CHAUVIN—CRO Agorapulse:
Sinon, c'est toujours un débat de quantité versus qualité : on en a fourni assez, oui, mais ils ne sont pas assez qualifiés. Tu recommences tous les soirs la même discussion. Donc si tu as quelqu'un qui est owner de la qualité et de la quantité en même temps, et qui porte ce message à toutes les étapes du funnel, c'est là que tu commences à avoir de la valeur.
Est-ce que j'étais spécialiste ? J'ai été BDR et GTM dans ma carrière, donc j'avais quand même une appétence côté business, mais je n'ai jamais porté un quota enterprise moi-même. Je vais être très claire : je n'avais pas cette compétence-là, et pourtant elle n'était pas nécessaire dans ce job.
Il y avait déjà des AE très bien, il y avait déjà des Head of Sales très bien aussi. En revanche, ce qui manquait, c'était vraiment cette capacité à faire un funnel de bout en bout qui fasse sens d'un point de vue argent, donc qui soit rentable, mais qui fasse aussi sens d'un point de vue développement. Sinon, on en revient à : j'ai pas assez de leads.

host—JORDAN CHENEVIER-TRUCHET:
Ouais, ok. Et du coup, toi tu arrives pour réconcilier deux mondes qui, dans beaucoup d'organisations, sont assez difficilement réconciliables. C'est quoi les premières étapes que tu as mises en place pour régler ce problème, à l'intersection entre le marketing et les sales ?

Le shift PLG vers mid-market : par où commencer

guest—SELMA CHAUVIN—CRO Agorapulse:
Je rajoute un petit élément de contexte : on partait d'un monde PLG transformé en un monde mid-market. Qu'est-ce qui se passe dans les modèles PLG ? Ça fonctionne 100% inbound, tu as des leads qui arrivent, tu les traites. Donc la première façon dont ils ont essayé de traiter le mid-market avant mon arrivée, c'est de se dire : on a juste des leads plus gros qui arrivent, on met un commercial en face pour les traiter.
Sauf que qu'est-ce qui se passe quand tu fais ça ? Déjà, tes budgets vont vite se retrouver bloqués, parce qu'acquérir des leads de qualité sur des boîtes assez grosses, soit tu te prends un mur, soit tu te prépares 18 mois en avance : on le sait, ça coûte très cher. Premier problème.
Deuxième problème : une fois que tes commerciaux sont habitués à ne vivre que sur de l'inbound, ils perdent leur capacité à générer eux-mêmes de l'outbound, du pipe. Troisième chose : le marketing continuait à être indexé sur sa capacité à faire du lead, et le sales sur sa capacité à closer.
Il n'y avait pas ce lien entre les deux. La première chose que j'ai faite, comme souvent quand j'arrive dans une entreprise, c'est de créer une équipe BDR au milieu, avec pour mission : je récupère les leads et j'en sors des rendez-vous.

host—JORDAN CHENEVIER-TRUCHET:
Cette équipe n'existait pas.

guest—SELMA CHAUVIN—CRO Agorapulse:
Elle n'existait pas. Ils ont essayé deux fois, ils se sont cassé les dents deux fois avant mon arrivée, parce que l'attente qu'ils avaient du BDR n'était pas réaliste : ils pensaient qu'un bon BDR était capable de faire du pipe from scratch. Tu donnes une liste de comptes et vas-y, démerde-toi. Ça fait à peu près dix ans que ça ne marche plus, je pense qu'on est d'accord.
L'idée, c'était plutôt de se dire : la première chose qu'on a, c'est un fort volume d'inbound marketing, donc poussons ça au maximum et voyons ce qu'on peut en tirer. Et ça, ça a fait notre croissance pendant un an. C'est juste cette capacité à mieux gérer les leads marketing, à les transformer via BDR puis via AE, qui nous a permis de faire 40% de croissance tout au long de l'année. Ce n'est pas négligeable.
Là, maintenant, on prend le deuxième effet kiss cool : je pense qu'on est un peu au maximum du ratio entre ce que j'investis et ce que je peux espérer en lead court terme. Il faudrait qu'on investisse fortement sur la marque et la création d'audience en amont, mais on sait que ça va prendre du temps. En attendant, il faut former les AE et les BDR à faire du vrai outbound, mais pas de l'outbound bête : de l'outbound intelligent, un outbound marketing.
Et c'est là que je me suis rendu compte que je ne maîtrisais absolument pas l'outbound, parce que c'est une tactique commerciale de base, et en fait les sales non plus. La seule façon de le faire, c'est via le marketing.

host—JORDAN CHENEVIER-TRUCHET:
100% d'accord.

guest—SELMA CHAUVIN—CRO Agorapulse:
J'ai un exemple tout bête. J'ai posé la question à mon patron commercial, parce qu'il me disait que l'outbound ne marchait pas. OK très bien, dis-moi ce que tu fais. Ah bah on prend des listes et on les appelle, mais on n'a rien à leur dire. Jusque-là, ça me paraît plutôt normal. Je lui ai dit : est-ce que tu essaierais pas quelque chose de tout bête ?
Tu tapes le nom des concurrents dans LinkedIn, il y a forcément quelqu'un qui va poster un post un peu énervé parce qu'ils ont fait une connerie. Tu regardes tous ceux qui ont liké, tu regardes les auteurs du post et tu les contactes. Vous avez fait ça ? Ah non.
OK. Est-ce que vous ajoutez naturellement sur LinkedIn toutes les personnes à qui vous parlez en réunion commerciale ? Parce qu'elles ont forcément des réseaux, et quand elles vont liker vos posts, LinkedIn va les proposer à d'autres personnes du même réseau. Nos AE n'ont pas ce réflexe. Les gars, on vend une solution qui gère les réseaux sociaux, est-ce qu'on pourrait être bons dans ce qu'on fait ?
Donc on a cette notion de l'outbound côté sales versus l'outbound côté marketing, qui pense déjà en système, au concept de quelle histoire on va raconter à qui. Ça a énormément de valeur, en vrai.

host—JORDAN CHENEVIER-TRUCHET:
Oui, oui. Proposition de valeur, audience, personnes. OK, d'accord.

guest—SELMA CHAUVIN—CRO Agorapulse:
On revient à la base.

host—JORDAN CHENEVIER-TRUCHET:
Ouais, ok. Ça m'a évoqué plusieurs questions. La première chose, c'est : toi, j'ai bien compris, tu as une équipe revenue. Cette question ne s'applique pas particulièrement à toi, mais je sais que beaucoup d'organisations se la posent, et ça m'intéresse d'avoir ton avis sur le sujet. Les BDR, dans ce cas-là, on voit bien qu'ils sont pile poil à l'intersection du marketing et des sales. Donc se pose régulièrement le débat : est-ce qu'on les met côté sales, ou est-ce qu'on les met côté marketing ? Vu que c'est des gens qui prennent le téléphone, la logique voudrait qu'ils soient côté sales. Pour autant, côté marketing, c'est particulièrement intéressant parce que ça leur permet d'être au contact des sujets de proposition de valeur, etc. Du coup, tu conseilles quoi ?

BDR côté sales ou côté marketing ? Sa réponse (et pourquoi elle change)

guest—SELMA CHAUVIN—CRO Agorapulse:
Alors j'ai pris des décisions très différentes, et là je vais te dire ce que je viens de décider : les BDR vont reporter, à partir du 1er juillet, à mon organisation commerciale — plus au marketing.

host—JORDAN CHENEVIER-TRUCHET:
OK.

guest—SELMA CHAUVIN—CRO Agorapulse:
Pourquoi ? C'est la vraie question. Par le passé, j'avais l'habitude de mettre les BDR au marketing. Pourquoi ? Parce que mon problème principal, c'était le traitement des leads marketing. Aujourd'hui, ce traitement-là, il est compris, il est intégré, ils savent le faire.
En revanche, là où ils ne sont pas bons, c'est quand le lead devient de plus en plus froid. C'est-à-dire que quand il n'y a pas assez de critères en amont — et aujourd'hui encore, je n'ai pas les volumes pour les nourrir de leads très chauds — j'ai besoin qu'ils soient meilleurs sur le traitement du froid, du tiède.
Et là, pour le coup, c'est une vraie compétence commerciale que de créer son portefeuille, de le travailler en profondeur. Un truc tout bête : je vais les mettre sur l'expansion. Il y a plein de trucs extrêmement évidents qu'on ne faisait pas, genre tu finis un client, tu vas taper dans toutes les autres filiales.
J'ai besoin que ce soit une compétence commerciale à la base : comment tu choisis, comment tu sais à qui tu vas parler, comment tu utilises la relation que tu as construite avec ton prospect devenu client pour aller démarcher les autres. J'avais besoin qu'ils acquièrent tout ça.
Donc demain, ils sont côté commerce. Ce qui ne veut pas dire que dans un an je ne vais pas changer d'avis.

host—JORDAN CHENEVIER-TRUCHET:
Et pourquoi des BDR pour gérer l'expansion, et pas les AE, alors ?

guest—SELMA CHAUVIN—CRO Agorapulse:
Parce qu'on est mid-market SMB, ce qui veut dire qu'il y a beaucoup, beaucoup de petites filiales et de petits comptes.

host—JORDAN CHENEVIER-TRUCHET:
OK.

guest—SELMA CHAUVIN—CRO Agorapulse:
Et que ça me coûte trop cher.

host—JORDAN CHENEVIER-TRUCHET:
D'accord, je comprends. Puis du coup, il faut développer la relation, mais ce n'est pas non plus le même type de relation que dans une grosse entreprise.

guest—SELMA CHAUVIN—CRO Agorapulse:
Et puis ce n'est pas la même personne. Si c'est la même personne qui gère, on est d'accord, c'est un commercial, tout va bien. Mais c'est vraiment : j'ai signé Décathlon France, je vais chercher Décathlon Allemagne. Décathlon a l'air d'être un mauvais exemple, parce que c'est une grosse boîte.

host—JORDAN CHENEVIER-TRUCHET:
Et en fait, j'allais te faire la remarque...

guest—SELMA CHAUVIN—CRO Agorapulse:
Ouais, mais en fait non. Quand tu te mets à l'échelle d'un département social media, c'est pas si...

host—JORDAN CHENEVIER-TRUCHET:
Gros. Oui, ok, intéressant.

host—JORDAN CHENEVIER-TRUCHET:
Et j'avais une autre question qui m'était venue sur les BDR.

guest—SELMA CHAUVIN—CRO Agorapulse:
C'est pratique.

host—JORDAN CHENEVIER-TRUCHET:
Oui oui oui, on en profite.

guest—SELMA CHAUVIN—CRO Agorapulse:
Et entre les deux, je reporte en direct, si ça t'intéresse.

host—JORDAN CHENEVIER-TRUCHET:
Entre les deux, quoi ?

guest—SELMA CHAUVIN—CRO Agorapulse:
Ils me reportent en direct.

host—JORDAN CHENEVIER-TRUCHET:
Les BDR.

guest—SELMA CHAUVIN—CRO Agorapulse:
Avant que je les mette aux sales.

host—JORDAN CHENEVIER-TRUCHET:
Quand ils étaient côté marketing, ils reportaient en direct.

guest—SELMA CHAUVIN—CRO Agorapulse:
Alors je reviens : quand j'ai créé l'équipe, je l'ai prise en direct, je l'ai mise au début ni côté marketing, ni côté sales.

host—JORDAN CHENEVIER-TRUCHET:
D'accord.

guest—SELMA CHAUVIN—CRO Agorapulse:
Et la raison pour laquelle j'ai fait ça, c'est que j'avais besoin de savoir où était le problème. Comme ils avaient essayé deux fois de monter cette équipe, et que deux fois ils s'étaient plantés, forcément, quand tu la recrées, tu as un peu de stress à te dire : je vais refaire la même connerie que ceux qui sont passés avant.
L'avantage, c'est qu'en les ayant en report direct, je pouvais voir les problèmes. Ça m'a permis d'être au contact du terrain et de spotter : ah ok, ça, ils ne savent pas traiter. Ah ok, ça, ils ne savent pas faire. Et de pouvoir les aider.

host—JORDAN CHENEVIER-TRUCHET:
Bon, très bien. Sur la logique du shift PLG vers mid-market : c'est quoi aujourd'hui les efforts que vous continuez de mettre sur la partie PLG ? Plus du tout ?

guest—SELMA CHAUVIN—CRO Agorapulse:
OK, en fait c'est assez naturel. Je vais te raconter une petite histoire. Quand j'étais chez DocuSign, j'étais en charge de la partie PLG. Je n'avais pas de budget, c'était : merci, est-ce que tu peux développer l'Europe pour nous ? Qu'est-ce que je suis allée faire ? Je suis allée prendre les budgets digitaux de tous mes collègues qui étaient sur la partie enterprise et mid-market, et je leur ai dit : je gère le digital pour vous, parce que je savais que neuf leads sur dix que j'allais générer seraient pour moi.
Donc j'ai pris leurs budgets, je leur ai donné le volume de leads dont ils avaient besoin, et tout le reste était pour moi. La même logique s'applique ici : une immense partie de ce qu'on fait ramène quand même du lead destiné au PLG. Et l'essentiel du travail, quand tu montes en taille de compte, c'est justement d'avoir une machine de triage pour ce que tu donnes à tes AE, et d'arrêter de passer du temps sur des outils qui ne sont pas du tout rentables.

Négocier un budget brand avec un CFO orienté court terme

host—JORDAN CHENEVIER-TRUCHET:
Oui, ok, je me souviens de ma question. Tu as dit que tu allais commencer à investir sur la partie notoriété, pour justement ouvrir un peu plus le scope de l'audience.
Comment tu arrives à faire ça, et notamment à convaincre de ça ? Parce que souvent, dans une boîte comme la vôtre, qui est très lean, où il y a des vrais enjeux de rentabilité, sur laquelle tu as déjà fait un shift PLG vers SLG, il y a vraiment cette logique de machine bien huilée que toutes les boîtes rêvent d'avoir.

guest—SELMA CHAUVIN—CRO Agorapulse:
Comment tu travailles ?

host—JORDAN CHENEVIER-TRUCHET:
Non mais voilà, tu vois, théoriquement en tout cas. Là, se mettre à investir dans la notoriété : moi, je comprends l'intérêt, mais est-ce que ton CFO comprend l'intérêt ? C'est-à-dire qu'à un moment donné, il va falloir lui expliquer : je vais investir plus d'argent, je ne vais pas récolter tout de suite, donc ça risque de dégrader mon CAC, etc.
Comment tu arrives à mener cette négo ?

guest—SELMA CHAUVIN—CRO Agorapulse:
En fait, on l'a géré on the side. Qu'est-ce qu'on a fait ? On a pris le budget destiné à la ligne classique, celui en rapport avec l'objectif du mois, du quarter, de l'année. On a fait notre budget, et ensuite on s'est mis une enveloppe à part dont on n'attend rien : on assume qu'elle n'est pas prise en compte quand on regarde les chiffres, et qu'on n'en attend aucun résultat avant six mois.

host—JORDAN CHENEVIER-TRUCHET:
Ah ouais ? Oui, bon, par contre, tu es tombée sur le bon interlocuteur pour accepter ce truc-là.

guest—SELMA CHAUVIN—CRO Agorapulse:
Mais en fait...

host—JORDAN CHENEVIER-TRUCHET:
Non.

guest—SELMA CHAUVIN—CRO Agorapulse:
Reprenons. Tu commences cette discussion en disant : voilà où en est mon CAC aujourd'hui, il ne va faire que se dégrader, parce que je vois que le coût d'acquisition d'un lead supplémentaire est très au-delà de mon coût d'acquisition lead moyen — ce qui veut dire qu'on a atteint un palier.
Donc soit on continue comme ça, et je sais qu'on ne va pas réussir à développer les revenus l'année prochaine, soit on accepte d'avoir ce budget on the side : comme ça, il ne dégrade pas les résultats de l'année en cours, mais on sait que ça nous permettra de garder le même niveau d'investissement l'année d'après, pour des résultats en hausse.
C'est un peu comme si je préemptais le budget 2027 en le pré-négociant dès 2026, parce que je sais que j'ai minimum un an avant de voir du résultat.

host—JORDAN CHENEVIER-TRUCHET:
Oui mais là tu es en train de me donner tes arguments de négo.

guest—SELMA CHAUVIN—CRO Agorapulse:
Ben oui, je ne suis pas sûre que...

host—JORDAN CHENEVIER-TRUCHET:
Oui, oui, non, non. OK. Et donc, tu arrives à convaincre de ça. Évidemment que c'est logique, mais comptablement, ton budget, il existe, et il a un impact sur la bottom line. Donc à un moment donné, il faut que tu arrives à faire comprendre ça aux gens.

guest—SELMA CHAUVIN—CRO Agorapulse:
Ailleurs dans le P&L, c'est tout. Je vais t'expliquer : j'ai un CEO qui est, pour le coup, un peu le CEO inverse de beaucoup : mon CEO est très, très marque, très, très dans ce genre de choses. Tu connais Aymeric, et on lance un nouveau produit dans quinze jours.
Ce nouveau produit, on sait qu'on a d'abord besoin de le confronter au marché. Ce que fait le produit, très vite, c'est Archi : l'idée, c'est que si tu prends une grosse pièce de contenu, par exemple ton webinar, il va t'en faire des shorts, des messages que tu postes sur les réseaux sociaux, en fonction de qui va poster, de ce qui marche mieux à telle heure, etc. Tout ce que tu peux imaginer.
Sauf qu'on sait qu'avant de pouvoir vendre ce produit pour de vrai, on a besoin d'avoir des retours de nos potentiels clients. Et on a surtout besoin de faire du volume, parce qu'encore une fois, c'est la tarification token. On sait tous que quand tu passes de la tarification au siège à la tarification au token, c'est hyper galère et casse-gueule. Donc on a besoin de cette phase de transition.

host—JORDAN CHENEVIER-TRUCHET:
Oui, tu as des conversations que t'as pas envie d'avoir ?

guest—SELMA CHAUVIN—CRO Agorapulse:
Exactement. Je suis absolument incapable de dire quel revenu ça va générer en 2026. Parce qu'entre le temps qu'on décide, le temps qu'on met en place, et le temps qu'on peut commencer à vendre, on est déjà en fin d'année. Sauf que qu'est-ce qui se passe si je ne le promeus jamais ?
Eh bien il ne marchera jamais, donc on n'aura jamais de clients avec qui construire. C'est dans ce cadre-là que j'ai réussi à expliquer ce besoin d'un budget on the side, qui n'est pas pris en compte dans le P&L mais qui sert les efforts long terme. Et une fois que la porte était ouverte avec ce produit, ça a ouvert la deuxième porte : faisons ça pour toutes les opérations de communication long terme.

host—JORDAN CHENEVIER-TRUCHET:
OK, et ta roadmap produit, tu la connais combien de temps à l'avance ?

guest—SELMA CHAUVIN—CRO Agorapulse:
On a des cycles de deux mois côté produit : on verrouille deux mois à l'avance ce qui va être produit. Et sinon, on a une idée vague de ce qui va sortir après.

host—JORDAN CHENEVIER-TRUCHET:
Ouais, mais du coup, tu peux capitaliser sur le fait que tu sais qu'on va sortir des trucs tous les deux mois. Tu ne sais pas exactement avec quelle avance, mais tu peux déjà préparer le terrain.

guest—SELMA CHAUVIN—CRO Agorapulse:
Ouais, exactement. Je vois beaucoup d'histoires de gens qui se mettent la rate au court-bouillon en pensant que la vraie date de lancement du produit doit être la vraie date de communication sur le marché. Mais en vrai, on s'en...

host—JORDAN CHENEVIER-TRUCHET:
Fout complètement.

guest—SELMA CHAUVIN—CRO Agorapulse:
Complètement. Les gens ne comprennent pas ça : le produit ne m'a pas dit quand ce serait dispo, donc je ne sais pas quand communiquer. J'en sais rien, je mets une date au pif dans trois mois, qu'il soit prêt ou pas prêt.

host—JORDAN CHENEVIER-TRUCHET:
Ah mais complètement, oui, oui, oui, 100%.
OK. Et je reviens du coup sur le switch PLG vers SLG. Ce qui se passe, c'est que généralement, quand on fait ce switch-là, on augmente le nombre de personnes dans les équipes, avec des BDR, etc. Vous, dans votre cas, vous avez fait ce switch-là, et en plus, la taille de l'équipe au global a été réduite.

Réduire l'équipe et dérégionaliser : tout en anglais, localiser le meilleur

host—JORDAN CHENEVIER-TRUCHET:
Qu'est-ce que vous avez fait ?

guest—SELMA CHAUVIN—CRO Agorapulse:
On l'a fait en deux temps. Quand je suis arrivée, ça ne fonctionnait pas, mais j'étais incapable de savoir si ce move vers le mid-market ne fonctionnait pas à cause de nous ou à cause du marché. Toujours le même problème. Donc quand tu regardes les trois choses à faire, dans l'absolu : déjà, il faut de la croissance ; ensuite, il faut que cette croissance soit scalable, donc être capable de montrer que tu peux faire ça à grande échelle ; et ensuite, il faut montrer qu'elle est profitable.
J'ai fait ces étapes dans l'ordre. Au début, j'ai staffé : pendant les six premiers mois, on a accepté et validé qu'on allait perdre de l'argent, le temps de staffer toute l'équipe et de voir ce qu'on était capable de générer en revenus. On a vu qu'avec cette méthode, on était capable d'augmenter fortement l'argent qu'on ramenait. C'était toujours pas dingue, mais c'était quand même du plus 30, plus 35%.
Une fois qu'on s'est dit ça, on a identifié qu'il y avait énormément de moyens investis dans des endroits qui n'arrivaient pas à décoller. Je donne un exemple : le marché allemand. On avait une personne qui faisait du contenu en allemand, une personne qui faisait du field en allemand, une personne qui faisait BDR allemand, un senior BDR pour la gestion des opportunités SMB, et un AE. Tout ça pour amener grosso modo 2 000 balles de MRR. Autant te dire que cette équipe était très, très loin de se payer.
Donc au début, on ouvre les vannes, on teste. Au bout de six mois, on relève les compteurs : qu'est-ce qui marche, qu'est-ce qui ne marche pas ? Je me suis rendu compte que mon budget marketing était beaucoup trop gras. Je l'ai divisé par deux. Ça m'a fait mal, parce qu'en ancienne CMO, je sais ce que ça fait de voir son budget divisé par deux. J'ai eu avec moi-même la discussion que je n'aurais jamais voulu avoir.
Deuxième chose : j'ai commencé une dérégionalisation. On avait ce playbook qui disait que pour être bon en Allemagne, il faut que tu fasses de l'allemand. J'ai assumé d'être mauvaise en Allemagne, tout simplement parce que je n'arrive pas à être bonne de toute façon. Donc acceptons d'être mauvais, et attrapons uniquement les personnes qui sont à l'aise avec le fait d'avoir des deals avec des équipes multiculturelles. Donc tout en anglais.
L'idée, au lieu de se faire chier à avoir en même temps tout ce qui sort en 50 langues, c'est de tout faire en anglais. Et les trucs qui marchent bien, on les localise : en français, ça marche toujours bien ; en allemand, ça marche toujours bien ; on ne localise pas beaucoup plus que ça.
Mais qu'est-ce que ça veut dire ? Ça veut dire que tu vas forcément avoir moins de pipe sur l'Allemagne pure, ou moins de pipe sur la France pure. Mais tu vas avoir plus de pipe worldwide, parce que tes contenus peuvent toucher plus de gens qui te coûtent moins cher. Un lead en Pologne, ça te coûte dix fois moins cher qu'un lead au UK, très clairement.
Sauf que quand tu fais ça, tu as quand même un allemand à nourrir. Donc on est sur la fin du process : les personnes qui sont mono-langue n'ont pas de place chez nous. C'est minimum ta langue plus l'anglais. On répartit tous les leads anglophones, et on te complète avec les leads qui tombent sur ta langue.
Ça aussi, c'est un gros travail dans la tête des gens. Parce que quand tu es seul sur un marché, c'est très territoire : ça, c'est mon territoire, c'est ma liste de comptes. Et non, moi, c'est non.

host—JORDAN CHENEVIER-TRUCHET:
Mais du coup, en plus de ça, la personne qui gère le marché allemand, elle est basée à Paris ? Ah, ok, d'accord, c'est un peu l'avantage tout de même : physiquement, tu n'es pas dans le pays que tu adresses, donc penser global est beaucoup plus logique. Quand tu es basée en Allemagne et qu'on te dit qu'il faut penser global, mais qu'en même temps on attend de toi que tu... il y a un truc un peu schizophrénique là.
Donc effectivement, il y a vraiment cette vision globale, avec ce truc où, pour les leads allemands, on attend de toi que tu puisses les traiter en allemand, par exemple.

guest—SELMA CHAUVIN—CRO Agorapulse:
Exactement. Mais dès le début, vu qu'on a cette politique de recrutement worldwide, par exemple, mon AE allemand, il est en Italie. Ça te verrouille déjà de très bons profils, parce que c'est difficile de trouver des jobs 100% remote quand tu travailles sur un autre marché que le tien. Mais surtout, ça nous permet aussi d'avoir quelqu'un qui est par défaut multiculturel — et donc ouvert à ça.

host—JORDAN CHENEVIER-TRUCHET:
Et oui.

guest—SELMA CHAUVIN—CRO Agorapulse:
C'est un bon moyen.

host—JORDAN CHENEVIER-TRUCHET:
OK, très bien. Tu as aussi fait un switch intéressant sur la manière d'envisager les différents pôles de l'équipe. Il y a les pôles classiques, surtout côté marketing, où tu pourrais avoir ton VP, ton CMO, ton machin-bidule. Là, tu as transformé ton équipe en raisonnant en termes d'étapes du funnel. Je veux bien que tu m'expliques ça.

Réorganiser par étape du funnel, plus par fonction

guest—SELMA CHAUVIN—CRO Agorapulse:
Alors déjà, ce qu'il faut savoir, c'est que ce qui coûte le plus cher dans une équipe revenue, c'est le management. Et maintenant que je le vois, ça me fait chier, parce que ce sont plein de gens qui coûtent cher — des managers — sans revenu associé en face.
Il faut être sûr que l'impact soit vraiment de la valeur. Le pire qui peut arriver, c'est de s'attaquer à ton CMO, puis ton VP marketing, puis ton Head of Demand Gen, puis ton équipe. Non seulement c'est chiant d'un point de vue argent, mais c'est aussi chiant d'un point de vue résultat, parce que moi, je n'ai absolument aucune visibilité de ce qui se passe. La chaîne est très lourde. Ça fait sens quand tu as de très grosses équipes, mais chez nous, quand je suis arrivée, il y avait 21 personnes en marketing, ça ne faisait pas sens. Surtout que maintenant, l'équipe, c'est plutôt dix ou douze personnes, donc absolument inutile.
Ce que j'ai fait, c'est d'avoir des patrons opérationnels en charge à 100% d'une étape du funnel, mais qui doivent aussi créer leur propre acquisition dans le waterfall à l'échelle. Tu as une équipe Brand et Content, dont l'objectif, c'est de générer de l'audience, de créer de l'attention.
Tu as une équipe Demand Gen — ce que j'appelle Go to Market, en fait un mix entre Demand Gen et PMM, on en reparlera plus tard si tu veux — qui est vraiment en charge de la conversion, mais qui est aussi en charge de certains sujets d'audience. Par exemple, c'est à eux de trouver des audiences pour faire du content syndication, avec un objectif de conversion, mais pas que : il y en a aussi avec juste un objectif d'audience.
Chacun a donc un objectif de création de contact, pour être sûr que les gens ne se posent pas uniquement en mode : j'attends que l'équipe d'au-dessus me donne tout ce dont j'ai besoin pour nourrir l'équipe d'après. Donc je crée une audience, je crée des leads, je crée des MQL ; les BDR les récupèrent et les convertissent en opportunité commerciale ; et les sales récupèrent ces opportunités et les convertissent en client signé.
Il n'y a plus cette notion de "je suis marketing versus je suis sales" et on se tape dessus, puisqu'il n'y a plus d'équipe qui porte ces notions-là. Il y a encore des histoires, ça n'a pas enlevé les histoires, elles sont juste plus petites.

host—JORDAN CHENEVIER-TRUCHET:
J'allais poser la question.

guest—SELMA CHAUVIN—CRO Agorapulse:
Non, mais ne rêve pas. C'est beaucoup plus gérable pour moi, parce que je suis manager direct de ces équipes : je peux apporter le même niveau d'attention à chaque étape du funnel, régler les bisbilles en étant là, tout simplement. Je n'ai pas à attendre que ça escalade et devienne ingérable avant que ça me touche.

host—JORDAN CHENEVIER-TRUCHET:
Bien sûr, bien sûr. OK, très très intéressant. Du coup, les métriques : la première partie du funnel, c'est la création d'audience, c'est quoi la métrique de succès pour ça ?

guest—SELMA CHAUVIN—CRO Agorapulse:
Pour moi, c'est uniquement un pool de comptes et un pool de gens que je vais pouvoir retravailler par la suite. C'est uniquement la pénétration de marché : combien de comptes ont interagi avec nous, d'une façon ou d'une autre, dans les douze derniers mois.

host—JORDAN CHENEVIER-TRUCHET:
Donc ça peut être de l'engagement social media.

guest—SELMA CHAUVIN—CRO Agorapulse:
Ça peut être de l'engagement social media, ça peut être des followers, ça peut être des likes, ça peut être n'importe quoi. Moi, j'ai un pool de comptes, et je veux savoir dans ces comptes-là combien ont commencé à nous connaître.

host—JORDAN CHENEVIER-TRUCHET:
D'accord. Donc à un moment donné, il y a quand même un travail de réconciliation de la donnée : mon follower, mon engagement, c'est potentiellement la même personne.

guest—SELMA CHAUVIN—CRO Agorapulse:
En gros, reprenons. Imagine que j'ai 20 000 comptes cibles worldwide. Mon travail, c'est de me dire : sur ces 20 000, combien ont commencé à interagir avec nous ? Peut-être que là-dedans, il y en a 10 000 qui ont réagi d'une façon ou une autre à n'importe quelle communication au cours de l'année. Ça, pour moi, c'est mon audience.
Et cette première équipe est évaluée sur sa capacité à faire augmenter ce chiffre, dans le pool que j'ai donné. La deuxième équipe, c'est sa capacité à ne pas uniquement faire de l'interaction, mais de l'interaction suffisamment qualitative pour être considérée prête pour un appel de BDR.
Donc ils ont un travail de conversion de cette audience, tout ce qui est nettoyage de données, etc. Mais ils ont aussi un travail de génération : des personnes déjà prêtes à être converties, parce que tout le monde n'attend pas que tu lui parles, il y a déjà des gens prêts à acheter.
Et donc ils ont ce double travail : travailler ce qui leur arrive de l'amont, et générer leur propre volume de MQL. Leur objectif, c'est vraiment la création de MQL en respectant les taux de conversion suivants. Chacun sait : mon objectif, c'est le volume que je dois donner à l'équipe d'après, et le respect des taux de conversion. Sinon, c'est trop facile.

host—JORDAN CHENEVIER-TRUCHET:
Non mais c'est sûr !

guest—SELMA CHAUVIN—CRO Agorapulse:
Voilà, donc ce volume de MQL avec ces taux de conversion : si les taux de conversion se dégradent, tu dois en fournir plus pour qu'au final on arrive au même résultat. Les BDR, c'est pareil : ils ont un objectif de production de rendez-vous, sur ce qu'apporte le marketing — donc l'inbound — soit sur ce qu'on appelle l'outbound, ou self-gen chez nous, pour ne pas dire auto-génération. Et les sales doivent traiter ce qu'on leur donne, mais aussi générer leur propre pipeline. On part du principe que chacun doit générer 20 à 30% de son résultat.

host—JORDAN CHENEVIER-TRUCHET:
D'accord.

guest—SELMA CHAUVIN—CRO Agorapulse:
C'est-à-dire que quand j'ai fixé des objectifs, je me suis assurée que les objectifs ne reposent pas uniquement sur l'activité de la personne en amont. Je donne un exemple : je suis ok si mon BDR ne traite que les leads marketing, il va réussir à faire son chiffre. Mais s'il veut faire son quota, il faut qu'il fasse de l'outbound.
S'il veut gagner de l'argent, il n'a pas le choix, il faut qu'il mette sa pierre à l'édifice. Donc je suis à peu près sécurisée, parce que le niveau de risque est relativement bas : je l'étale sur tout le monde, et j'enlève une partie de la construction de pipe de chacun. Donc si mon système fonctionne depuis le début, ça passe. Mais pour que chacun soit performant, il faut qu'il réussisse cette partie perso.

host—JORDAN CHENEVIER-TRUCHET:
Oui, et le fait de réussir cette partie perso, en plus, ça te permet de te dire : si jamais le pipe qu'on est censé m'amener en amont ne vient pas, j'ai les compétences pour compenser. Il y a aussi ce truc-là.

guest—SELMA CHAUVIN—CRO Agorapulse:
Ces discussions typiques que j'ai avec le patron commercial, c'est que chaque semaine, il commence en disant : j'ai pas assez de pipe, j'ai pas assez de demandes de démo. Je suis ok, merci de me donner l'info, mais tu en es où de l'outbound ? On va commencer par cette discussion-là.
Non, mais l'outbound, on n'y arrive pas. Ah, mais tu comprends, c'est difficile, parce que la demande de démo, c'est plus facile, tu crois ?

host—JORDAN CHENEVIER-TRUCHET:
Oui, et du coup, ça, c'est ta responsabilité, et l'autre, c'est la responsabilité des autres.

guest—SELMA CHAUVIN—CRO Agorapulse:
Et à chaque fois, c'est : oui, mais pour qu'on réussisse, il faut qu'untel fasse ça. Sinon, il y a des choses que tu peux contrôler. Donc une fois que tu es parfait sur ce que tu peux contrôler, là tu as le droit de venir te plaindre de ceux d'avant dans la boucle.

host—JORDAN CHENEVIER-TRUCHET:
Mais oui.

guest—SELMA CHAUVIN—CRO Agorapulse:
Mais c'est tout. Et c'est bête, mais ça change tout.

host—JORDAN CHENEVIER-TRUCHET:
Bah, c'est clair, oui. Au moins, tout le monde est au clair sur ses responsabilités.

guest—SELMA CHAUVIN—CRO Agorapulse:
OK, tu fais ton chiffre, c'est ta responsabilité. Tu n'as pas assez de demandes de démo ? On s'en fout, tu es censé faire 50 à 60% avec ta propre prospection.

host—JORDAN CHENEVIER-TRUCHET:
OK. C'est quoi votre CRM ?

guest—SELMA CHAUVIN—CRO Agorapulse:
Alors on est en pleine migration. Aujourd'hui, on a encore Salesforce, et on est en train d'en sortir.

host—JORDAN CHENEVIER-TRUCHET:
OK, ça nécessite quand même deux choses, un peu techniques : tes 20 000 comptes, ça te demande au départ de bien les mapper, d'identifier qui sont les personnes à l'intérieur de ces comptes qu'on va adresser, s'il y a différentes filiales, etc.

guest—SELMA CHAUVIN—CRO Agorapulse:
Non, même pas.

host—JORDAN CHENEVIER-TRUCHET:
Même pas cette partie-là ?

guest—SELMA CHAUVIN—CRO Agorapulse:
Non, même pas. OK, des grosses feignasses, déjà. Et puis on a faim. Et ça, ça demande beaucoup, encore une fois : est-ce que c'est rentable de faire tout ce travail ? Je ne suis pas méga convaincue.

host—JORDAN CHENEVIER-TRUCHET:
Oui, mais oui. Et puis, en fait, le truc, c'est que quand le compte arrive, ton BDR regarde s'il y a des filiales, et le sujet est réglé.

guest—SELMA CHAUVIN—CRO Agorapulse:
Enfin, si on se dit qu'il faut 15 000 comptes touchés, que je les mappe avant ou que je regarde après, pour 50 000 €, je m'en fous en fait. Une fois que j'en ai 15 000, c'est bon.

host—JORDAN CHENEVIER-TRUCHET:
Oui, oui, non, on est d'accord. Mais donc, c'est 20 000 comptes cibles : tu vas quand même faire le travail en amont de te dire qui ils sont, pour ensuite travailler sur la mesure de l'engagement, etc. Tout ça, tu le fais dans ton CRM, et ça nécessite aussi que tes BDR, dans leur travail de prospection, etc.
Il y a vraiment un travail là. De toute façon, j'imagine que ton CRM devient vraiment le centre névralgique, là où tu sais que tu n'es pas en train d'appeler une personne qui est par ailleurs déjà travaillée par quelqu'un d'autre. D'accord, ok, bon, très bien. Tu as mentionné...

Go-to-market : ce que ça veut vraiment dire vs. ce que tout le monde en fait

host—JORDAN CHENEVIER-TRUCHET:
Go to market : les personnes en charge du go to market, qu'est-ce que tu mets dans la définition de go to market ?

guest—SELMA CHAUVIN—CRO Agorapulse:
Alors à la base, pour qu'un programme marche, pour que n'importe quoi marche, ce qu'on se dit toujours, c'est qu'il faut des gens qui vont définir l'audience à qui tu vas parler, le message phare, et tout le PMM. Ensuite, il y a le contenu, qui va faire la grosse pièce de contenu qui raconte l'histoire qu'on vient de définir. Et ensuite, tu as la demande.
Leur travail à eux, qu'est-ce que c'est ? C'est de trouver des canaux sur lesquels diffuser tout ça. Sauf qu'une étape qu'on oublie énormément, c'est que juste trouver un canal, ça ne marche pas tant que ça. Ce qui a de la valeur, c'est le canal à l'intersection de l'audience, de la façon de consommer, des habitudes, etc.
Et on néglige encore plus ça. On bidouille, parce qu'on n'est pas habitué à faire tout ça. Donc on fait un énorme travail en amont, avec PMM et content, mais ensuite, côté Demand Gen : si tu veux du trafic, démerde-toi, fais une campagne LinkedIn. Ça ne marche pas comme ça.
Donc ce que j'ai fait, c'est que j'ai mis Demand Gen chez le PMM, et j'ai appelé ça Go to Market. En mode : les gars, vous me faites des recommandations sur l'histoire, c'est très très bien. Maintenant, je veux des recommandations sur les canaux, sur les influenceurs, sur quelle newsletter sponsoriser. Et vous fournissez ça en même temps que votre analyse de marché.
Et le rôle de Demand Gen, c'est de discuter avec vous pour choisir, dans tout ce qu'ils vous donnent, et d'ajouter bien sûr ce à quoi ils pensent, mais de choisir la bonne intersection : canal, histoire et ciblage, on va dire.

host—JORDAN CHENEVIER-TRUCHET:
Ouais, parce que je pose la question parce que je vois de plus en plus le terme go to market devenir important. Avant, j'avais pas l'impression qu'on utilisait autant l'expression go to market, et maintenant on l'utilise vraiment à toutes les sauces.
Je pense que beaucoup de gens l'emploient sans vraiment réfléchir à ce que ça veut vraiment dire, le go to market. C'est pour ça que je me permets de poser la question. Et souvent, j'ai le sentiment que ce que la plupart des gens appellent aujourd'hui go to market, c'est en fait la fonction revenue : à la fois le marketing et les sales.
Et donc go to market, ce serait un peu cette espèce de truc global qui génère du revenu, non ? Toi, là, tu es vraiment sur, littéralement, la mise sur le marché.

guest—SELMA CHAUVIN—CRO Agorapulse:
Oui, oui, c'est vraiment ça. Non, c'est vraiment très, très concret.

host—JORDAN CHENEVIER-TRUCHET:
Mais oui, oui, oui.

guest—SELMA CHAUVIN—CRO Agorapulse:
Comment on va au marché avec le produit qu'on doit vendre ?

host—JORDAN CHENEVIER-TRUCHET:
Oui, totalement. Mais du coup, moi, dans ma position — tu vois, prestataire de services, etc. — il faut que je m'adapte au vocabulaire de mes clients. C'est-à-dire qu'un client va utiliser une expression, et je vais avoir la même expression chez quelqu'un d'autre, mais ça ne voudra pas dire la même chose.

guest—SELMA CHAUVIN—CRO Agorapulse:
Tu vois GTM Engineer dans la boucle ?

host—JORDAN CHENEVIER-TRUCHET:
Exactement, exactement. Des GTM Engineer qui s'occupent beaucoup de faire de la demande, tu vois. Donc en fait...

guest—SELMA CHAUVIN—CRO Agorapulse:
C'est du RevOps. Et du coup, on appelle ça GTM Engineer.

host—JORDAN CHENEVIER-TRUCHET:
Exactement. Et donc le truc, c'est qu'on avait eu cette discussion avec mon co-fondateur, Enguerrand : à un moment donné, on s'est dit, mais les gens, ils veulent dire quoi par go to market ? Parce que go to market, littéralement, c'est mise sur le marché, donc c'est effectivement ça, j'imagine.
Mais on se rendait bien compte qu'il y avait un problème : on n'était pas en train de parler de la même chose. Et maintenant, effectivement, il y a les GTM Engineer et tout. Bon, bref. Mais ça permet de bien recadrer, parce que je pense qu'il y a des gens qui vont entendre ce terme et ne pas y aller pareil. Tu as raison.

guest—SELMA CHAUVIN—CRO Agorapulse:
Surtout qu'avant, j'avais beaucoup le discours "j'ai pas assez de lead" de la part des CEO. Maintenant, c'est "j'ai un mauvais GTM". En fait, ils veulent dire la même chose : j'ai pas assez de lead.

host—JORDAN CHENEVIER-TRUCHET:
Exactement, exactement, mais avec un terme un peu plus technique.

guest—SELMA CHAUVIN—CRO Agorapulse:
Ils ont l'impression que du coup, ils ont maîtrisé le sujet marketing.

host—JORDAN CHENEVIER-TRUCHET:
Ouais, complètement. OK, intéressant. Et donc effectivement, chacune des équipes a des objectifs. En fait, toutes les équipes ont un objectif de taux de conversion. C'est juste que la première équipe a également un objectif de volume.

guest—SELMA CHAUVIN—CRO Agorapulse:
Alors, c'est que tout le monde a un domaine qui est chasse gardée, mais en plus, ils sont en charge de générer leur propre business.

host—JORDAN CHENEVIER-TRUCHET:
Donc il y a quand même une partie de volume et de conversion dans tous les...

guest—SELMA CHAUVIN—CRO Agorapulse:
Cas, en tous les cas.

host—JORDAN CHENEVIER-TRUCHET:
OK, intéressant. J'ai envie qu'on parle un peu de l'IA, qui est le nouveau sujet du moment — j'en ai forcément entendu parler. Qu'est-ce que vous faites avec l'IA ?

L'IA en système : agents, feedback loops et scoring automatisé

guest—SELMA CHAUVIN—CRO Agorapulse:
Alors, c'est sûr que c'est juste ça la question, parce que la réponse va être super longue.

host—JORDAN CHENEVIER-TRUCHET:
Non, mais bon, évidemment, vous en mettez un peu partout. Alors je vais poser la question différemment : qu'est-ce que vous avez mis en place, et sur lequel tu te dis, ça, c'est vraiment le truc qui a un vrai gros impact, qui est transformatif pour la boîte ?

guest—SELMA CHAUVIN—CRO Agorapulse:
J'ai envie de te dire, c'est une philosophie : c'est de l'avoir pensé en système. Ce que je vois beaucoup, c'est des gens hyper hypés qui vont tenter un petit POC, sans jamais le scaler. Et le truc va vivoter tout seul dans son coin, il va mourir de sa belle mort quand quelqu'un arrêtera de le mettre à jour.
Sauf que nous, ce qu'on aime faire, c'est l'inverse. On a regardé tout le funnel, on a regardé toutes les étapes qui marchaient mal, et celles qu'il était trop coûteux de corriger avec de l'humain. Désolée, je parle vachement argent.

host—JORDAN CHENEVIER-TRUCHET:
Mais c'est très...

guest—SELMA CHAUVIN—CRO Agorapulse:
Bien, il faut que ce soit rentable, ce qu'on fait. Donc on a regardé toutes ces étapes-là, et on a vu des éléments... par exemple, c'est stupide, mais quand tu montes en taille de compte, tu as forcément beaucoup plus de besoins légaux, tu commences à recevoir des RFP. Est-ce que ça vaut le coup d'embaucher un juriste qui va traiter trois deals dans le mois, ou d'avoir une armée prête pour les fois où ton panier moyen explose ? Non.
Donc ce qu'on s'est dit, c'est ça : par exemple, on a un agent légal, on a un agent de réponse à RFP. On a construit comme ça des petits Husky sur tous les endroits où on avait vraiment besoin de ressources, de bras en plus, et où on n'avait pas les compétences, ou pas le temps.
C'était la première étape. Une fois qu'on a construit tous ces petits agents qui vivaient les uns à côté des autres, un peu comme tout le monde, on s'est posé la question : comment tu les maintiens ? Et donc trouver un owner pour ces sujets-là, et ensuite comment tu en fais un vrai système, et tu sors des petits agents créés sur Claude ou sur ce qui te fait plaisir, pour avoir une vraie logique. C'est l'étape qu'on est en train de boucler : on est sortis des POC, on a commencé l'industrialisation.
Je donne un exemple tout bête : le lead scoring. Aujourd'hui, tu as des MQL dans ta base de données, tu as des BDR qui les traitent. Comment tu fais dans la plupart des boîtes ? Tu dis au BDR : celui-là, faut que tu l'appelles, tu appelles ça comme tu veux, tu fais du scoring, peu importe.
Sauf que là-dedans, il y a un truc qu'on oublie : chaque jour, il se passe quelque chose dans l'entreprise de certains leads. Pour plein de raisons — d'un coup, il y en a trois qui ont cliqué sur un contenu. Le BDR accepte ou refuse ce lead, et son comportement va driver ce qu'on va lui envoyer par la suite. Plus tu acceptes de leads qui ressemblent à ça, plus Claude va se dire : ce sont des bons leads, donc j'ai intérêt à en envoyer plus qui leur ressemblent. Donc on se sort de toute la partie scoring, hyper chiante à maintenir, avec juste un système de : je prends, je ne prends pas.

host—JORDAN CHENEVIER-TRUCHET:
C'est le swipe Tinder, quoi.

guest—SELMA CHAUVIN—CRO Agorapulse:
Exactement, non mais c'est exactement ça. Et on assume, on dit : ne le prends pas si tu n'en veux pas, parce qu'on va t'en donner plus des pareils. Donc à toi de choisir.

host—JORDAN CHENEVIER-TRUCHET:
Ok, et moi j'ai l'impression qu'il y a un système de feedback loop. Mais cette feedback loop, à un moment donné, il faut qu'elle ait un impact sur les personnes en entrée.

guest—SELMA CHAUVIN—CRO Agorapulse:
Justement : si, par exemple, l'équipe Demand Gen est évaluée sur le volume de MQL qui réussissent à convertir, et que les BDR les renvoient — donc ce ne sont plus des bons MQL — ça ne les intéresse pas de générer ce type de lead. Donc ils vont créer plus de campagnes qui donnent le bon type de lead.

host—JORDAN CHENEVIER-TRUCHET:
D'accord.

guest—SELMA CHAUVIN—CRO Agorapulse:
OK, et ça marche pour toutes les étapes. C'est pareil entre la Brand et la conversion. Si les pièces de contenu n'ont pas généré les volumes attendus, la Brand sait qu'elle doit les changer, parce qu'eux, ils ont un volume à faire sur le mois. C'est leur problème.
Je donne un exemple tout bête : ils m'ont donné le plan de Q3, je regarde le plan, et je vois quelque chose qui est quand même très bas de tunnel. Je leur dis : les gars, à mon avis, ça ne va pas vous faire le volume. Ça va vous mettre à mal pour l'ensemble du quarter, parce qu'en plus, vous l'avez créé en début de quarter.
Est-ce qu'on n'y a pas pensé ? Oui, mais vous avez une logique de volume, donc si vous voulez le faire, allez-y. Prévoyez un plan B qui dit : si on n'a pas le volume, ou si ça ne convertit pas mieux que le reste, on fait autre chose. Une feuille de papier comme ça.

host—JORDAN CHENEVIER-TRUCHET:
Exactement, c'est ce que j'allais te...

guest—SELMA CHAUVIN—CRO Agorapulse:
Dire. Mais ça ne convertit pas mieux que le reste, il faut que vous me proposiez autre chose. Et tu vois, cette façon de raisonner, elle n'est pas naturelle encore.

host—JORDAN CHENEVIER-TRUCHET:
Ouais, ouais, ok. Bon, par...

guest—SELMA CHAUVIN—CRO Agorapulse:
Contre. Du coup, on sort un peu de l'IA. Maintenant qu'on a la première étape — les agents — on a l'IA dans ce côté feedback loop : maintenant que j'utilise ce truc, quel impact ça a sur l'équipe qui est en amont de moi ?
La deuxième étape, en parallèle, ça demande un gros travail sur la data : le cleaning de data, le dashboarding. Chez nous, on a un peu le problème du riche, c'est qu'on a 50 millions de dashboards partout, parce que la facilité de faire des dashboards, ça a fini par ça. Et on le sait. Donc maintenant, je suis en mode méta-dashboard : maintenant que vous avez tous ces beaux dashboards hyper sympa de votre côté, est-ce que je pourrais tout rapatrier, s'il vous plaît ?
Donc voilà où on en est. C'est pas ultra fancy, c'est pas le use case qui fait rêver. De notre côté, c'est vraiment très méthodo, et on a réglé des problèmes. Les problèmes sont réglés, c'est comment tu t'assures que le problème n'arrive plus.
Un exemple tout bête : l'outil RFP. Maintenant qu'on a un outil RFP qui sait écarter les mauvais, comment on s'assure de ne récupérer que les bons en entrée ? Quels sont les critères qui font que c'est un bon RFP qu'on va recevoir ? Donc on aura moins besoin d'utiliser cet outil, précisément.

host—JORDAN CHENEVIER-TRUCHET:
Ouais, ok. Et donc, la réponse au RFP est aussi aidée par l'IA. Oui, en fait, j'ai l'impression aussi que quand je prends tes différents use case — RFP et légal — c'est vraiment des Husky parfaits : c'est une compétence spécifique qui prend beaucoup de temps à traiter. L'IA, c'est juste parfait pour ça. Tu déroules ton agent sur cette compétence, et il te réduit le temps. C'est un truc de dingue, en fait.

guest—SELMA CHAUVIN—CRO Agorapulse:
Et avant, j'avais une petite crainte : personne n'est là en backup. On a un avocat, bien sûr, mais personne n'est vraiment là en backup. En fait, on n'a rien. OK, quel est le risque en vrai ? S'il y a un problème, qu'est-ce qui se passe ? C'est pas énorme. OK, donc je prends le risque. Quand j'ai vraiment un doute, j'appelle l'avocat, mais 90% des cas sont traités.

host—JORDAN CHENEVIER-TRUCHET:
Mais bien sûr. Donc oui, après, c'est le coût du risque, voilà. Mais ça se calcule comme un bénéfice.

guest—SELMA CHAUVIN—CRO Agorapulse:
Le prochain outil qu'on met en place au 1er juillet, c'est un outil qui va faire de la démo automatisée. Est-ce que ça veut dire que je ne veux plus qu'il y ait de démo faite par les sales ? Pas du tout. Mais je pense que pour 80% des deals — encore une fois, on est passé de PLG à in-market — c'est ultra transactionnel, et tu n'as pas besoin d'intelligence humaine qui coûte une blinde.
Et encore une fois, il y a de la valeur là-dedans. Si tu mets des sales à traiter des deals à 200 balles, le sales va être extrêmement énervé, il va se barrer, parce qu'il n'est pas payé pour faire ça, et lui-même te coûte trop cher pour ce que c'est. Donc tu es un peu embêté, parce que dès qu'il en signe plus, tu ne t'y retrouves pas, et ça t'empêche de libérer du temps pour aller sur le haut de segment. Tu te contentes d'une petite croissance qui ne t'arrange pas.

host—JORDAN CHENEVIER-TRUCHET:
Et pour autant, tu sais qu'il y a besoin de ce contact, de cette démo, sinon la personne ne convertit pas. Pas exactement, c'est pile poil le truc entre les deux.

guest—SELMA CHAUVIN—CRO Agorapulse:
Là, d'où le truc : il va y avoir cet outil de démo guidée. En revanche, on a mappé tout le parcours idéal, et tous les points de drop, où c'est ok, un humain reprend la main. Est-ce qu'on lance ça ? Au début, en vrai, ça va nous enlever quoi, peut-être 10 à 20% des deals ? Mais c'est pas grave.
Une fois qu'on aura compris... quel est le problème à cet endroit-là, là où ça drop, on remet toujours un BDR. Qu'est-ce qu'on peut faire pour limiter la friction ? Et on va progresser étape par étape. Une fois qu'on a fait les étapes, feedback loop : si c'est toujours ce pattern de gens qui coincent à cette étape-là, est-ce qu'on en veut, ou pas, comme client ? Non ? Alors on arrête les campagnes qui ramènent ces gens-là.

host—JORDAN CHENEVIER-TRUCHET:
Oui, on en revient à ce qu'on disait tout à l'heure. Cette phase de démo, c'est vraiment une construction très product, comme un onboarding produit, où tu te dis il faut que je retienne les gens. Est-ce que l'équipe Product est impliquée dans cette partie-là ?

guest—SELMA CHAUVIN—CRO Agorapulse:
Alors oui, en partie. C'est notre équipe solutions consulting qui gère cet outil. Elle est déjà en relation quotidienne avec le produit pour travailler l'environnement de démo, les différents pitchs, etc.

host—JORDAN CHENEVIER-TRUCHET:
Ouais, mais en même temps, l'IA te permet de ne pas prendre trop de temps à l'équipe product, parce que tu vas pouvoir sortir des écrans rapidement.

guest—SELMA CHAUVIN—CRO Agorapulse:
En fait, on a trouvé un outil : tu le briefes, et il se promène dans ta plateforme. Tu lui donnes accès à tout ce que tu veux, il se promène vraiment dedans, il clique, il donne un avis sur les onglets. Tu n'as pas besoin de faire des vidéos, des screenshots, rien. Il se débrouille.

host—JORDAN CHENEVIER-TRUCHET:
C'est quoi le nom du produit ?

guest—SELMA CHAUVIN—CRO Agorapulse:
Le nom du produit... le nom du produit, lequel on a choisi, je ne sais plus, ni comment. Car Oui.

host—JORDAN CHENEVIER-TRUCHET:
Car Oui, ok, j'en ai entendu parler.

guest—SELMA CHAUVIN—CRO Agorapulse:
Ben, on écoute, on teste. Et bien, tu pourras l'utiliser.

host—JORDAN CHENEVIER-TRUCHET:
Exactement, top, ok. Et donc, moi, j'ai l'impression que vos cas d'usage sont très spécifiques : ça ne part pas dans tous les sens, ça règle un problème précis.
Je reviens ici sur un sujet. En fait, j'ai pensé à ça au moment où tu as parlé de dashboard...

Le piège des dashboards sans contexte humain

host—JORDAN CHENEVIER-TRUCHET:
Un risque que j'ai déjà vu dans pas mal d'équipes, c'est le fait de trop faire confiance à son IA.

guest—SELMA CHAUVIN—CRO Agorapulse:
Oh que oui !

host—JORDAN CHENEVIER-TRUCHET:
Donc tu sais déjà où je vais. J'ai fait le lien avec les dashboards : le nombre de boîtes où, quand tu as ton dashboard, tu as la data, tu penses qu'elle est correcte, et puis tu vas vérifier à la main, et tu te rends compte que ton dashboard te dit n'importe quoi. Ça arrive dans beaucoup d'équipes.
C'est quoi le process que vous employez, dans la mise en place de l'IA, pour éviter ce genre de trucs ?

guest—SELMA CHAUVIN—CRO Agorapulse:
Alors, ce dont je me rends compte aujourd'hui, c'est que quand l'IA commence à déconner, c'est parce qu'on l'a mal nourrie en amont, ou qu'on a laissé de l'imprécision s'installer.

host—JORDAN CHENEVIER-TRUCHET:
Tout à...

guest—SELMA CHAUVIN—CRO Agorapulse:
Fait, ça c'est clair. L'exemple typique, parce qu'on le fait tous les jours : on va créer une page Notion pour le lancement d'un projet, avec l'IA. On ne va pas la corriger parfaitement, il va rester deux, trois erreurs qui traînent. Sauf que ce doc va servir à briefer l'IA sur un autre projet, et au fur et à mesure, les problèmes grossissent. Au final, on se retrouve avec un truc qui n'a rien à voir avec ce que tu voulais faire à la base.
Même chose sur les dashboards : un chiffre faux, il va s'en servir comme base et l'appliquer partout. Donc le secret, pour moi, il est multiple : un, les sources de données ; deux, à qui tu donnes accès à ce doc.
Première chose, les sources de données : je le fais à l'envers, c'est-à-dire que les docs que l'IA va utiliser sont faits par des humains. Un exemple tout bête : mes équipes commerciales me fournissent, tous les mois, un MBR — Monthly Business Review — avec les performances de chaque AE, les stats de chacun, etc. Ils peuvent utiliser l'IA pour créer ce document, mais ils le revoient à la main. Et je veux un commentaire humain sur chacune des performances.
C'est ce que j'ai utilisé pour mon dashboard, parce que ce doc a eu le stamp humain, donc je suis ok pour m'en servir. Et c'est ça pour tout : je ne veux pas de l'extraction Salesforce en bulk. Si je veux de la data, je veux qu'elle soit dans un doc qui peut être rempli via Claude et Salesforce, mais vérifié par un humain, avant d'aller dans le méta-dashboard que j'utilise au quotidien. Et chaque personne est responsable des données qui arrivent jusqu'à moi.
Deuxième chose : à qui tu donnes accès à ce doc ? Parce que le problème, c'est que la plupart des gens ne savent pas lire. Je donne un exemple : j'avais un BDR, ce mois-ci, à zéro en pipeline généré. Évidemment, mon collègue RH est venu me dire : il est nul, il faut le virer. Je regarde bien dans le doc : là, il n'a rien de validé, mais il a 20K en pipe. C'est juste que les meetings n'ont pas encore eu lieu, parce que l'AE était absent la semaine dernière.
C'est tout bête, mais c'est beaucoup moins évident que ça en a l'air. Le premier truc que j'ai fait, quand on m'a dit qu'il était à zéro, c'est d'aller regarder ce qu'il avait en pipe. Mais ça, ce n'était pas dans le dashboard.

host—JORDAN CHENEVIER-TRUCHET:
Oui.

guest—SELMA CHAUVIN—CRO Agorapulse:
Donc si la personne n'est pas formée à la lecture de ce dashboard, et si ce n'est qu'un outil d'aide à la décision et pas un outil de décision, c'est là que tu te plantes.

host—JORDAN CHENEVIER-TRUCHET:
Oui, mais au final, c'est exactement le même sujet que l'IA. Ce dont ton RH avait besoin, c'était d'un contexte. Et ce dont l'IA a besoin pour performer, c'est d'un contexte.

guest—SELMA CHAUVIN—CRO Agorapulse:
Et le contexte doit rester humain, parce qu'il n'y a que l'humain qui a ce niveau de nuance. C'est pour ça que je te dis : quand je crée mes dashboards, c'est fait sur des documents avec du contexte donné par un humain. Je ne veux pas la stat du gars, je veux pourquoi c'est important ou pas. Je pense — ou alors c'est parce que je suis nulle — mais je ne veux pas faire l'impasse sur cette étape.

host—JORDAN CHENEVIER-TRUCHET:
OK, si je me fais un peu l'avocat du diable : tu n'as pas la garantie que la personne n'a pas mis un commentaire humain sans analyse. Tu peux très bien mettre un commentaire bête et méchant sur une donnée que te donne l'IA.

guest—SELMA CHAUVIN—CRO Agorapulse:
Surtout les juniors, on le sait tous : le slide avec le diagramme, "a augmenté de 40% ce mois-ci", point, voilà le commentaire.

host—JORDAN CHENEVIER-TRUCHET:
Oui, oui. Donc tu vois, le truc, c'est que je comprends l'intention, mais ça ne te garantit pas.

guest—SELMA CHAUVIN—CRO Agorapulse:
Pas du tout. Rien ne le garantit.

host—JORDAN CHENEVIER-TRUCHET:
On est d'accord. Voilà, c'est ça que je voulais t'entendre dire.

guest—SELMA CHAUVIN—CRO Agorapulse:
Non, il n'y a rien qui va garantir. La seule chose que ça garantit, c'est que quand j'ai ces infos, je peux utiliser mon esprit critique pour aller creuser.

host—JORDAN CHENEVIER-TRUCHET:
Oui, et puis c'est aussi qui porte ça. Voilà, j'ai l'exemple.

guest—SELMA CHAUVIN—CRO Agorapulse:
Un manager commercial : un de ses AE a fait zéro le mois dernier, il a mis une phrase de commentaire, et il a fait zéro. "Mais c'est pas grave." Je lui dis : non, ça, ça ne me va pas comme réponse, tu vas creuser pourquoi, et tu me confirmes.

host—JORDAN CHENEVIER-TRUCHET:
Donc tu es contre la déresponsabilisation ?

guest—SELMA CHAUVIN—CRO Agorapulse:
Oui, totalement. Sauf que dès que tu déresponsabilises, la stratégie et l'intelligence, tu trouves...

host—JORDAN CHENEVIER-TRUCHET:
Ben, justement.

guest—SELMA CHAUVIN—CRO Agorapulse:
C'est intéressant, mais c'est un point de vue totalement opposé.

host—JORDAN CHENEVIER-TRUCHET:
Mais c'est justement pour ça, tu vois. C'est un travail où cet esprit critique et d'analyse est particulièrement important, et à réaliser par des humains. Arriver à tirer le fil de ce qui a vraiment de la valeur, c'est un humain qui est capable de le faire, en ayant la connaissance de son audience. Et ça, l'IA...

guest—SELMA CHAUVIN—CRO Agorapulse:
Je vais te donner un exemple. Comment on a fait ? Vas-y, ce fameux...

host—JORDAN CHENEVIER-TRUCHET:
Produit à l'IA.

guest—SELMA CHAUVIN—CRO Agorapulse:
Je vais donner un exemple. Comment on a fait ?

host—JORDAN CHENEVIER-TRUCHET:
Vas-y, go !

guest—SELMA CHAUVIN—CRO Agorapulse:
Ce fameux produit qu'on lance dans quinze jours, comment est-ce qu'on l'a conçu ? Comment on l'a designé ? En fait, on a demandé à tous les AE de poser une question lors d'un call avec leur prospect. Une question toute bête : quelle est votre plus grosse problématique autour de la gestion de contenu aujourd'hui ?
Basique, comme question. Sauf que : combien de temps tu penses que ça mettrait d'analyser à la main des réponses dans plein de langues différentes, pas forcément structurées, pas forcément au fil de l'eau ? Nous, on a utilisé l'IA. Pourquoi ? Pour rassembler, collecter toute cette information, et faire des résumés.
Et le rôle du PM, c'est d'aller creuser quand il est alerté. 90%, c'est confirmatoire : ça va conforter des idées qu'on a déjà. Mais l'important, c'est : ok, lui, il n'a pas pensé pareil, pourquoi ? Tu veux écouter ce call-là pour comprendre le use case. Or, tu vas te rendre compte que sur ce vertical, tout le monde pense la même chose : tu en écoutes deux, trois, tu te dis, ok, a priori c'est cohérent, j'ai pas besoin d'y passer du temps.
Ce que fait l'IA, ça te permet d'identifier, dans l'immensité des choses que tu as à faire, les sujets qui méritent que tu y passes du temps. Sinon, tu traites tout, pareil, comme si on avait juste posé la question et écouté chaque réponse en consacrant autant de temps à chaque industrie. Sauf qu'en fait, il y a peut-être 2 à 3 industries où on a vraiment besoin de creuser.
Et typiquement, ce que ça nous a permis de sortir, ce sont 4 à 5 Husky. On les avait déjà, en vrai, donc ça nous a plutôt confortés sur ces 4 à 5 cases. Mais on s'est rendu compte qu'ils n'avaient pas la priorité qu'on pensait leur donner. Un exemple tout bête : le use case founder-led, où la personne doit poster en son nom et au nom de sa boîte sur LinkedIn. On l'avait priorisé bas, pour nous, c'était un petit use case. En fait, on s'est rendu compte que dans toutes les boîtes, les social media managers disaient : mon patron a ce problème.

guest—SELMA CHAUVIN—CRO Agorapulse:
Et du coup, on va peut-être le re-prioriser. Parce que s'ils ont tous ce problème de leur boss, et que c'est le boss qui a l'argent, ça vaut peut-être le coup qu'on en parle un peu plus.
Mais l'IA toute seule n'aurait pas pu spotter ça. Et l'humain tout seul non plus.

host—JORDAN CHENEVIER-TRUCHET:
Oui.
En fait, on pense la même chose. C'est juste qu'on ne positionne pas le rôle de l'IA au même endroit sur le use case PM. Moi, je suis 100% d'accord pour dire que la manipulation de la donnée et l'analyse de données, c'est un use case sur lequel l'IA est imbattable. Imbattable.

guest—SELMA CHAUVIN—CRO Agorapulse:
Tu ne peux pas lutter.

host—JORDAN CHENEVIER-TRUCHET:
Non, mais exactement. Je considère même que le fait qu'une partie de l'activité d'un PMM, ce soit de l'analyse de données, c'est presque pas normal. Ils doivent le faire, mais c'est une compétence tellement particulière... Du coup, c'est cool, maintenant j'ai l'IA qui me permet de faire ça, je peux me concentrer là où je vais être capable d'apporter de la valeur. Et on est d'accord que là où le PMM peut vraiment apporter de la valeur, l'IA n'en apporte pas vraiment.

guest—SELMA CHAUVIN—CRO Agorapulse:
Enfin, l'IA va t'apporter une valeur moyenne : la moyenne de tout ce que les gens qui ont posé ce type de question ont eu comme réponse. Donc tout dépend de ton attente. Je comprends très bien ceux qui se disent : ce n'est pas un sujet prioritaire, mais j'ai besoin d'occuper un peu l'espace, je vais faire du moyen, je vais mettre de l'IA dessus parce que j'ai pas le temps. Mais on en revient au même : ton arbitrage est juste plus haut, et tu te dis, ce sujet, je m'en fous.

host—JORDAN CHENEVIER-TRUCHET:
C'est un sujet d'arbitrage, on est clairement là-dedans.

guest—SELMA CHAUVIN—CRO Agorapulse:
C'est de la ressource, et aujourd'hui l'humain coûte plus cher que l'IA. Peut-être qu'on changera bientôt.

host—JORDAN CHENEVIER-TRUCHET:
Ah bah la semaine dernière, on a eu une petite fenêtre sur les nouveaux modèles qui vont bientôt coûter autant que l'humain. Tu te dis, bon, là, on va revenir sur des questions fondamentales.

guest—SELMA CHAUVIN—CRO Agorapulse:
Mais c'est ça. En fait, si j'ai ce volume de ressource, j'ai cet argent, je le mets où ? Qu'est-ce qui doit être automatisé, là où je peux gagner de l'argent, et qu'est-ce qui doit être fait par un humain ? Ce ne sont que des arbitrages.

host—JORDAN CHENEVIER-TRUCHET:
Oui, et d'ailleurs, c'est marrant qu'on parle de ça, parce qu'on enregistre au moment de VivaTech. Et un des sujets qui est revenu hier, c'est justement les nouveaux modèles qui sortent, Claude, des modèles très puissants, etc.
Et il y a aussi cette question du modèle économique de ces entreprises : à un moment donné, il faut prendre tout le marché, lever beaucoup d'argent, etc. On connaît. Mais tôt ou tard, il faut retrouver une équation économique.
Donc ces modèles, à un moment donné, vont se mettre à coûter plus cher. Et quand ton IA va en venir à te coûter autant que tes humains, je pense que les gens qui se disent aujourd'hui "vas-y, volume, volume, volume" vont revenir à des questions du genre : est-ce que je recrute vraiment cette personne-là ? Ce serait un peu le même genre de réflexion que tu pourrais avoir via l'usage de l'IA.

guest—SELMA CHAUVIN—CRO Agorapulse:
J'ai envie de nuancer un peu. Rappelle-toi, à l'époque des premiers ordinateurs — enfin, je ne t'ai pas connu à cette époque, c'est tout — mais en gros, avoir un peu de puissance de calcul en plus, ça coûtait une fortune. Maintenant, le prix du token...

host—JORDAN CHENEVIER-TRUCHET:
C'est vrai.

guest—SELMA CHAUVIN—CRO Agorapulse:
Non, mais c'est vrai qu'aujourd'hui, ça coûte cher parce qu'on utilise aussi les mauvais outils pour les mauvaises tâches. Typiquement, moi, je vois le modèle le plus cher réglé par défaut, parce que...

host—JORDAN CHENEVIER-TRUCHET:
Mais pareil.

guest—SELMA CHAUVIN—CRO Agorapulse:
Parce que je ne vois pas la différence.

host—JORDAN CHENEVIER-TRUCHET:
Et puis, de toute façon, dans le forfait max, tu n'atteins jamais la limite.

guest—SELMA CHAUVIN—CRO Agorapulse:
Dès que tu fais de l'image, mais bon. Mais ça passe, tu vois. Donc on n'a pas ce besoin de se restreindre. Si demain je me dis : ok, j'ai ce crédit, il faut vraiment que ça tienne jusqu'à 17h, parce que sinon à 15h j'en ai plus, alors là, tu vas commencer à utiliser différents modèles, c'est vrai.
Et le prix des petits modèles n'est pas forcément élevé, et ils suffisent pour la plupart des tâches dont tu as besoin. Par contre, gérer ta boîte mail, mettre à jour de la doc suite à des trucs que tu as faits, tu n'as peut-être pas besoin du modèle le plus puissant, en fait.

host—JORDAN CHENEVIER-TRUCHET:
Complètement. Et d'ailleurs, je pense que la majorité des tâches marketing, d'une manière générale, ne nécessitent pas les meilleurs modèles. Je pense vraiment.

guest—SELMA CHAUVIN—CRO Agorapulse:
On va arriver à un nouveau truc où, après avoir dit que le marketing n'était pas un vrai métier parce que c'était du bon sens, maintenant on va dire que ça ne prend pas beaucoup de crédit IA, donc ça n'a pas de valeur.

host—JORDAN CHENEVIER-TRUCHET:
Honnêtement, non. Mais voilà, c'est aussi un peu l'opportunité de se dire : en réalité, ce qu'on produit, ce n'est pas comme si on devait faire de la recherche fondamentale, par exemple.

guest—SELMA CHAUVIN—CRO Agorapulse:
La puissance de calcul, non.

host—JORDAN CHENEVIER-TRUCHET:
Tu vois. Enfin bon, ok, j'ai envie de finir sur un sujet un peu d'ouverture...

Concilier CRO, famille et compétition sportive internationale

host—JORDAN CHENEVIER-TRUCHET:
Dans la préparation de cet épisode, tu m'as parlé un peu à côté du fait que tu essaies de concilier ce nouveau rôle — encore assez nouveau, on est d'accord —, ta vie de famille, et ton entraînement pour prétendre au sport à l'international. Comment tu fais pour gérer tout ça ? C'est quoi ton secret de productivité ?

guest—SELMA CHAUVIN—CRO Agorapulse:
Je pense que ça peut arriver à beaucoup de monde : si on te dit que l'imprévu, c'est horrible, c'est vraiment toi. Je ne dirais pas non.
Je m'entraîne cinq fois par semaine, mine de rien : chaque fois, c'est 1h30 à 2h, donc il faut que ça rentre dans un emploi du temps, comme tout le monde, j'ai envie de dire. Si j'ai 1h à midi, je vais réussir à caser 45 minutes d'entraînement, ça fera toujours ça de moins à faire.
Le soir, je vais m'entraîner souvent : je mets les enfants à table et je vais m'entraîner, parce qu'il n'y a pas le choix, faut que ça rentre. Ou alors je mange avec eux et je m'entraîne après. Ou alors je me réveille plus tôt et je m'entraîne avant le week-end. Ils savent qu'en général, tout ce que je n'ai pas pu faire dans la semaine, je le fais le week-end, donc maman n'est pas là avant trois ou quatre heures.
Ce qui a changé ma vie, c'est que je me suis fait une salle de sport chez moi, parce que déjà...

host—JORDAN CHENEVIER-TRUCHET:
J'allais te poser la question, parce que je me dis, mais attends, j'y...

guest—SELMA CHAUVIN—CRO Agorapulse:
Arrivais pas là, c'est chez moi.

host—JORDAN CHENEVIER-TRUCHET:
Maintenant. D'accord.

guest—SELMA CHAUVIN—CRO Agorapulse:
Mais au début, je n'avais rien comme matériel : j'avais une barbell et quelques poids, c'était très roots. Maintenant, j'ai une vraie salle où je peux m'entraîner, sécurisée, etc. Par étapes.

host—JORDAN CHENEVIER-TRUCHET:
Ouais. Et donc tu as un coach qui vient à la maison ?

guest—SELMA CHAUVIN—CRO Agorapulse:
Alors j'ai deux coachs, parce que je fais deux sports à l'international : l'haltérophilie et la force athlétique. Et d'ailleurs, c'est marrant, parce que la semaine prochaine, j'ai les championnats d'Europe d'haltérophilie et les championnats de France de force athlétique la même semaine. Je suis extrêmement énervée.

host—JORDAN CHENEVIER-TRUCHET:
Ah ouais, c'est pas mal, tu vas finir ta semaine...

guest—SELMA CHAUVIN—CRO Agorapulse:
Sur les rotules, c'est clair. J'ai des coachs qui me suivent en vidéo. Pour l'haltérophilie, je vais dans un club, parfois le soir quand je peux. Sinon, mon coach vient à la maison. Pour la force athlétique, c'est à distance, et il vient avec moi seulement sur les compétitions.

host—JORDAN CHENEVIER-TRUCHET:
OK, et du coup, on commence à se lancer sur ce sujet-là : la sécurité. Parce que tu fais des sports où, généralement, il y a des gens qui veillent au grain. Toi, tu es chez toi, tu gères. OK, j'ai mis les enfants à manger, et là, je me fais coincer sous la barre : il se passe quoi ?

guest—SELMA CHAUVIN—CRO Agorapulse:
C'est pas possible, parce que le matériel est sécurisé. Un exemple tout bête : le banc. Quand tu fais du développé-couché, mon banc a des sécurités sur les côtés, qui font que si la barre tombe, elle reste au-dessus de mon cou. Ce sont des bancs de compétition.
Pareil pour le squat : j'ai une cage à squat avec des montants, donc même si je m'effondre, la barre reste dessus, moi je suis en dessous, en sécurité. L'haltérophilie, c'est un peu plus dur, parce que tu n'as pas de cage.
Mais j'ai, par exemple, des plots : quand je fais des mouvements lourds en hauteur, si la barre tombe, elle tombe sur les plots, elle ne m'écrase pas. Le risque zéro n'existe pas, mais j'ai mon téléphone pas loin, et j'ai de la sécurité.

host—JORDAN CHENEVIER-TRUCHET:
Et puis c'est toujours une affaire d'arbitrage, on en revient toujours à la...

guest—SELMA CHAUVIN—CRO Agorapulse:
Norme. Et tu peux te casser une jambe en courant, tu sais.

host—JORDAN CHENEVIER-TRUCHET:
Non, mais c'est vrai. OK, excellent. Mais donc, finalement, le hack, c'est juste de super bien organiser sa journée, de...

guest—SELMA CHAUVIN—CRO Agorapulse:
Fuir les imprévus comme la peste. Je pense que tous les parents ont ça, tu fuis les imprévus. Déjà, les gens qui t'appellent entre 18h et 20h, on en parle ? Le tunnel du soir, je préfère prendre des réunions après 20h, une fois que j'ai géré les enfants. Mais dans l'intervalle, vous m'oubliez, je ne suis pas joignable.
Il y a aussi : qu'est-ce que tu fais si, par exemple, je me qualifie pour les championnats du monde la semaine où j'ai mon off-site avec tout le Comex ? Je fais quoi pour manger ? Parce que qui dit pré-compét dit nourriture de pré-compét.
Donc est-ce que j'assume d'être la chiante qui dit non, je ne mange pas ce qui est prévu, aussi fancy soit-il, et je ramène toute ma nourriture ? Ou je trouve une façon de m'arranger sur place ? C'est plein d'arbitrages à faire, encore une fois.

host—JORDAN CHENEVIER-TRUCHET:
Je vois très bien, j'ai le même genre de problème. Ce week-end, je vais faire un col en vélo, mais c'est une semaine parisienne. Généralement, une semaine parisienne, c'est un moment où le soir, je vais boire des coups, etc. Là, il faut que je me tienne.
OK, super intéressant. Et tu arrives à dormir 8h par nuit ?

guest—SELMA CHAUVIN—CRO Agorapulse:
Ben oui, parce que l'avantage, c'est que tu dors très très bien. La fatigue, ça t'aide, oui.

host—JORDAN CHENEVIER-TRUCHET:
OK d'accord. Bon très cool.
Dans l'audience, il y a des gens qui font du marketing, tu adresses une audience qui fait du marketing. C'est aussi le moment de parler des prochaines étapes. Pour finir, sur Agorapulse, qu'est-ce qui va se passer prochainement ?

guest—SELMA CHAUVIN—CRO Agorapulse:
Donc, rappelons : Agorapulse, c'est une solution de gestion des réseaux sociaux. Ça veut dire que la personne a une problématique : pouvoir publier sur les réseaux sociaux en masse, répondre à tous les messages privés qu'elle reçoit via l'inbox, ou pouvoir reporter sur l'ensemble de son activité.
C'est ça qu'on fait aujourd'hui. On lance, comme je le disais, dans deux semaines, Archi — je ne sais pas quand l'épisode va sortir, il sera peut-être déjà sorti d'ici là. Donc Archi, c'est un produit qui répond à la problématique principale des boîtes mid-market aujourd'hui : elles ont plein de contenus dans tous les sens, elles ne savent jamais quoi en faire, ni comment l'utiliser au mieux, notamment sur les réseaux sociaux.
Ce qui se passe souvent, c'est que tu as ton pauvre gars des RS, et tout le monde lui envoie plein de pistes de contenu en disant "fais-moi un post, s'il te plaît" — et on sait que ça ne marche jamais. Donc l'idée, c'est que cette personne va pouvoir mettre dans Archi sa pièce de contenu — son podcast, par exemple, ou son ebook — et automatiquement, Archi va découper sa vidéo en shorts qu'elle pourra poster sur Insta, par exemple, va lui proposer des hooks, des approches pour parler de ce contenu de plein de façons différentes.
Il va rédiger lui-même les posts, en respectant son tone of voice et celui de l'entreprise. On va aussi être capable d'intégrer des notions de design, si on a rentré toutes les guidelines de l'entreprise. En gros, ça simplifie le délai et les prises de tête entre "j'ai produit ma grosse pièce de contenu" et "je vais vraiment l'exploiter au maximum sur le marché", au lieu de la publier une fois, envoyer un email, et passer à la suivante.

host—JORDAN CHENEVIER-TRUCHET:
OK. Et quand on parle de réseaux sociaux, c'est aussi bien les pages entreprises que les pages personnelles.

guest—SELMA CHAUVIN—CRO Agorapulse:
Tu peux avoir les deux.

host—JORDAN CHENEVIER-TRUCHET:
OK, merci Selma pour tout ça. Et pour les personnes qui veulent suivre ce que tu fais, tu t'es mise à poster beaucoup sur LinkedIn depuis... c'est très cool, en fait.

guest—SELMA CHAUVIN—CRO Agorapulse:
Je suis obligée, parce que c'est comme ça que je comprends les vraies problématiques de ma cible. Bien sûr, je ne le fais pas pour moi.

host—JORDAN CHENEVIER-TRUCHET:
Mais non, moi, quand j'ai vu que tu te mettais à publier beaucoup, je me suis dit : han han.

guest—SELMA CHAUVIN—CRO Agorapulse:
Elle a compris son métier.

host—JORDAN CHENEVIER-TRUCHET:
Non, mais c'est très cool. Merci d'avoir partagé tout ça. Pour les personnes qui veulent suivre ce que tu fais, on les envoie sur LinkedIn. Merci beaucoup.

guest—SELMA CHAUVIN—CRO Agorapulse:
Merci à toi.

Introduction — What is Loop Marketing?

host—JORDAN CHENEVIER-TRUCHET:
Hey Kipp, welcome to the show.

guest—KIPP BODNAR—CMO, HubSpot:
Hey Jordan, thanks for having me, man. Appreciate it.

host—JORDAN CHENEVIER-TRUCHET:
You're one of the people who helped popularize inbound marketing at HubSpot over a decade ago. Now you've written a book with Kieran Flanagan — a previous guest on the show — about what you call Loop Marketing. So what is Loop Marketing?

guest—KIPP BODNAR—CMO, HubSpot:
Loop Marketing is the answer to how the heck you do marketing in the post-AI world. Whether you love it or hate it, your buyers are adopting AI and using it, and that shows up in your business in a lot of ways. Your marketing numbers probably aren't as good, because search traffic is getting disrupted. More people are spending on paid ads, so paid costs go up. More companies are sending email, so open and click rates get unpredictable. When your reps get on a call, the buyer has already done way more of the discovery, because they've been going back and forth with an assistant. In that world you have to operate in a fundamentally different way.

guest—KIPP BODNAR—CMO, HubSpot:
Loop Marketing is a four-step framework for any marketing activity in the post-AI world. First, you Express the idea you have — AI makes creating content better and easier than ever if you have the right process and data. Then you Tailor it: how do you personalize it in email, in ads, in another channel, so it resonates with your target individual. Third is Amplify: how do you distribute that message — a lot has changed with creators, with answer engine optimization, with how you show up in AI search. And the last step is Evolve: how do you learn really quickly. Learning compounds, and AI makes it much faster. How have your metrics changed, what equations inform your strategy, and how does your team need to work? The fundamental way of operating has changed dramatically.

The ChatGPT moment and the broken funnel

host—JORDAN CHENEVIER-TRUCHET:
What was the trigger that made you realize the old funnel is dead and that we need to change the way we see inbound?

guest—KIPP BODNAR—CMO, HubSpot:
It goes back to the ChatGPT moment. When ChatGPT launches, it becomes a new way people access information, and search starts to change. Google starts changing its algorithm aggressively to prepare for the AI search shift, and traffic to websites starts to decline. Then you start seeing AI-generated video and ads, and the whole buyer technology adoption process changes — it felt like overnight. We famously went viral for losing millions of blog visits in a year because of all this disruption. When you're in the midst of that, the world is just a different place, and doing more of what we were doing was not going to make us successful.

HubSpot's strategic shift: creators, video, newsletters

host—JORDAN CHENEVIER-TRUCHET:
I remember a post from Kieran explaining that you invested much more into creators and into more dimensions. Can you walk me through the strategy?

guest—KIPP BODNAR—CMO, HubSpot:
When you lose a bunch of traffic, the easiest thing to say is "we'll just do more of what we did" — and that's not the right path. Before all this happened, we realized we had concentration risk: we were really dependent on Google search traffic. I'd never have predicted the speed and extent of the disruption, but we were thinking about it. So we bought The Hustle years ago. That was our first step toward saying the world is moving past static text-based blog content and much more toward multimedia.

guest—KIPP BODNAR—CMO, HubSpot:
Kieran had this great slide: you want to be part of your buyer's day. They search throughout the day, but they're also listening to podcasts, reading email newsletters, seeing ads, participating on social. You need to show up across those places. So we invested heavily in YouTube, in a network of newsletters for different profiles of our audience, and far more aggressively in brand and storytelling. There was a period where marketing was just knobs and dials — put money in Google and Facebook, do some programmatic SEO, and you know what you'll get. That's fundamentally not true anymore.

guest—KIPP BODNAR—CMO, HubSpot:
If I gave specific advice: video, and email newsletters whether on Substack or self-hosted. When I say video I largely mean YouTube, but extend it to TikTok, Reels and even LinkedIn short-form, depending on how consumer or B2B you are. We have a whole chapter on creators in the book — they're transforming how audiences are influenced and discover products. We partnered very heavily with creators and do more every month. The other side: we leaned into data and AI to personalize, and got big improvements in email and SMS, because AI helps create bespoke messages for each person.

host—JORDAN CHENEVIER-TRUCHET:
I watch a lot of Marketing Against the Grain — great show. And I see you partnering with shows like My First Million; every video, HubSpot is there.

guest—KIPP BODNAR—CMO, HubSpot:
My First Million is an interesting story. We acquired it when we bought The Hustle, and nobody cared about podcasts back then — it was almost a throw-in. I figured if we got that podcast as part of the acquisition, we could use it as the cornerstone to build out a network, because a cold start is really hard. It's been an awesome partnership with Sam and Sean. Then there are other podcasts we don't own but partner with deeply on creative and advertising.

Express: start from your point of view

host—JORDAN CHENEVIER-TRUCHET:
Back to the Loop framework — it starts with Express, which means starting from your unique point of view. Usually marketers start from what the buyer wants. So it's flipping it: you start by giving instead of taking. How does that look behind the scenes at HubSpot?

guest—KIPP BODNAR—CMO, HubSpot:
If you're a student of marketing history — you read Ogilvy, the old direct-mail copywriting books, a great book is "Different" by Youngme Moon — one core theme runs through all of it: differentiation. For most of marketing history, the differentiation lived in the story and in what you chose to focus on. There's the classic Ogilvy line: "at 60 miles an hour the loudest thing in this Rolls Royce is the clock." Every other carmaker focused on other benefits; he focused on luxury, quality and silence. We lost that in the last 20 years, because differentiation moved from story to tactics — if I was a little better at Meta ads, I could be good enough. AI is making that no longer true. You now have to be far more differentiated on the story side.

The yoga studio test and differentiation

guest—KIPP BODNAR—CMO, HubSpot:
How do we approach this at HubSpot? The easiest and most fun test — credit to our co-founder Brian Halligan — is: could a yoga studio say this? Put someone else's logo on your message; could it credibly come from them? It's basic but it works. Right now in AI everyone says the same thing: agents, headless, CLI, speeds and feeds — all this sameness. So I ask: what is everyone saying, and what's the exact opposite? Usually the opposite is interesting, and often it actually fits you if you change a few things. Then: can I deliver it in unique ways? In the inbound days we'd write posts about how direct mail was killing the environment — find every angle to be David versus Goliath. That got lost in the last 10–20 years, and I think we're getting back to it.

The white paper test

host—JORDAN CHENEVIER-TRUCHET:
I have a test I run with customers — the white paper test. Take your headline, especially the website one, where people try to be inspirational instead of just saying what they do. Put it on a blank sheet, show your neighbor and ask: what am I selling? And second: can you tell it's me and not another company? Same test as yours.

guest—KIPP BODNAR—CMO, HubSpot:
Quick detour: the website is the bane of every marketer's existence, because everyone has an opinion — legal, the CEO, everyone. There's the old Ogilvy line: there are no statues of committees. Remarkable work comes from a singular mind and a singular idea, and the more people get involved, the more watered down it gets. That's what happens to websites. Your test is a forcing function — go back to the group and say: we're trying to do everything; can we just do one thing very well?

host—JORDAN CHENEVIER-TRUCHET:
Sometimes you read a headline and don't even understand what the company does. People confuse a value proposition with a slogan. "Because you're worth it" is a great slogan — you could sell a mattress with it — but it's not a value proposition. That's where people get mixed up: they try to be inspirational and miss the pragmatism.

guest—KIPP BODNAR—CMO, HubSpot:
Humans are inherently self-interested, so they hide behind jargon, in-group colloquialisms and meaningless slogans, because being specific risks alienating someone they know. In some ways the best marketers are empathetic sociopaths: I understand what this person wants, and I genuinely don't care what the world thinks about it — is this the right answer? If so, do the thing. Most marketing work is a reflection of how sensitive the leaders of the company are.

The four Rs and new equations

host—JORDAN CHENEVIER-TRUCHET:
Once the point of view is defined, you give marketers the four Rs: remarkability, resonance, reach, regenerative. Walk us through them.

guest—KIPP BODNAR—CMO, HubSpot:
What we're trying to say in the book is that the rules of the game have changed. I live by a set of principles — one is: I never play a game I can't win. So who wins? Not the most skilled team — the people who understand the rules, and they're usually the people who write the rules. So you have to write the rules. What I mean is there are new equations for doing marketing well. An old equation was cost per lead: I advertise on Meta and care deeply about cost per lead. But what's the actual equation for an ad with a great creator who has a stranglehold of trust over an audience? I wouldn't measure them on cost per lead alone — I'd completely undervalue the relationship. I need a different equation.

guest—KIPP BODNAR—CMO, HubSpot:
The four Rs are the guardrails for the outcomes you're trying to achieve. Concretely: I value a creator on a composite — cost per lead, plus the impact on my AI visibility, plus the brand lift. I decide what to pay based on that. That gets me resonance with their audience. Then I let them build the message, not me, because I don't know their audience — they do. That makes it remarkable, because it's native. Then I look at what worked and didn't, give the data back so they can regenerate and run the next campaign better. These are post-AI constructs, and most companies aren't doing any of this yet.

From SEO to AEO

host—JORDAN CHENEVIER-TRUCHET:
Say I'm a CMO whose past growth came from SEO. I have a traditional SEO team, a blog factory, a content calendar. What should I change?

guest—KIPP BODNAR—CMO, HubSpot:
You have humans and money, so a couple of things. You've done text-based SEO — now you need text-based answer engine optimization, and you need to understand how visible you are across ChatGPT, Perplexity, Gemini and Claude. Get a visibility and recommendation tool — we have one at HubSpot, there are tons out there. SEO and AEO are a Venn diagram: you need authority and great content for both, but what those mean differs. Answer engines cite certain sources far more — LinkedIn, YouTube, Reddit — to build consensus. Where Google's PageRank asked "who has the most authority," answer engines ask "what's the consensus answer across the internet." So step one: add AEO. Don't abandon SEO, but move along the spectrum.

guest—KIPP BODNAR—CMO, HubSpot:
Step two: move from text-only to multimedia — at a minimum, text to video. Video's hard, and you only have so many people and dollars. So you need content automation tools that get the same SEO and AEO results with less writing and editing time, which frees people up to write scripts and make videos. YouTube is one of the biggest search engines on the planet, and video is a key component of both AEO and SEO. You're building a network of content that lets you be discovered wherever the buyer is.

Video and on-screen talent: how to convince people

host—JORDAN CHENEVIER-TRUCHET:
A lot of CMOs are convinced they need to invest in creators and video, but nobody around them — especially the CEO — wants to put their face online. They know it works but can't find the words to convince people. How do you get someone to start showing up?

guest—KIPP BODNAR—CMO, HubSpot:
It's hard, and it's situational. If it's a super technical business you need credible experts. There are a few paths. One: someone on YouTube in your market is already doing it — acquire them, an acqui-hire of their audience and channel. That's a different equation, because you have to believe that attention, influence and distribution are the scarce resources of the modern world. I believe it'll be far easier to commoditize product and features than to commoditize distribution and awareness.

guest—KIPP BODNAR—CMO, HubSpot:
Two: take one of your best people and say this is now your job. There aren't many great sales YouTubers, because the math doesn't work for them — they make a ton selling, and you won't pay them the same to be on YouTube. So sometimes you say: no, I am going to pay you the same, because this is existential to how we build community and inform customers. Three: find someone out in the world who could be good at it — they spoke at your events, did a few things — and turn it into a partnership, a contract, or a hire. Those are the three paths.

Creator ROI: the wrong question

host—JORDAN CHENEVIER-TRUCHET:
And the ROI? Most CEOs say: show me we can get leads from this.

guest—KIPP BODNAR—CMO, HubSpot:
There's the very basic math: I'll pay X for the person, Y for production and editing, I'll get Z views and convert maybe two percent into leads. That's the math most people would do, and I think it's pretty dumb. Don't confuse simple with similar — you can be simple and very differentiated. The biggest thing that equation misses: most companies with any demand motion are so constrained they're paying break-even return on ad spend just to keep volume going. So this becomes existential. Who gives a shit what the ROI is — I am not going to be a growing business if I don't do this. If I don't find a new channel of distribution, eventually I'm losing money on Google and Meta and never getting the volume I need. I have to make this work.

host—JORDAN CHENEVIER-TRUCHET:
Tough conversation for some CMOs next week, but an honest one.

Agent-first HubSpot: the platform

host—JORDAN CHENEVIER-TRUCHET:
Let's talk HubSpot. For the past two or three years you've repositioned around AI agents — customer agents, prospecting agents. What does "agent-first company" actually mean for HubSpot today?

guest—KIPP BODNAR—CMO, HubSpot:
A lot gets lost in the technical jargon — CLI, skills, all of it. Simply: we're building an agentic customer platform where you have all your customer data, with two modes of working. If you're technical, you build on our APIs — three layers. A classic data API for raw data. A context API with pre-computed, inferred context — for example, the key characteristics of your top 1,000 performing SDR emails — so you can build a skill off of it. And a work API where you call agents, skills and tools to conduct work autonomously. If you run RevOps or sales ops, you'll love working that way.

guest—KIPP BODNAR—CMO, HubSpot:
If you're non-technical — a rep or a marketer — you work with HubSpot three ways. First, you Ask: our Breeze assistant, which has gotten really good and I use every day, where you build artifacts and interact with all your customer context. Second, you vibe-code: build landing pages or a webinar page in an editor experience, even if you're non-technical. Third, you build custom agents with no code for your specific tasks. Technical people build on the APIs; everyone else works in those three ways, plus our premiere agents.

The terminal shift and multiplayer AI

host—JORDAN CHENEVIER-TRUCHET:
The way I work has changed dramatically in six months — I spend time in the terminal, which was never the case. We're a HubSpot customer, and now I work on HubSpot from my terminal: I create campaigns, manage contacts. So I spend less time on the HubSpot website but maybe more time doing things on HubSpot. Do you see this shift in the numbers, and how do you see its future?

guest—KIPP BODNAR—CMO, HubSpot:
The chasm of adoption is very wide. There are two main groups: people like you and me — terminal open, codex open, Breeze assistant, ten tasks running in parallel — and people who do nothing with AI. For that second group, even Breeze building a workflow is transformative. For the early minority, anyone working through a connector — Claude Code, OpenAI, ChatGPT, codex — uses the product more and gets more value. I don't think it matters where the work happens; what I care about is delivering the right outcome to you. If you want to run our prospecting agent out of Claude Code with our CLI, go nuts.

guest—KIPP BODNAR—CMO, HubSpot:
Any technology revolution starts with a big unbundling — lots of platforms and ways to work — and then things consolidate. Right now people should work wherever works best for them, which is why we built the API layer. The economics work for us either way, in-app or not. I'll be honest, though: Claude Code is not that fun — hitting allow all the time, downloading and uploading files, managing everything locally, a fragmented mobile-versus-desktop experience. It's the ultimate single-player experience; you're isolated from everybody. Those are real drawbacks if you run a real team. For collaboration, you and others will work more in HubSpot.

host—JORDAN CHENEVIER-TRUCHET:
Two comments. One: multiplayer is the next evolution of tools like Claude — everyone agrees. Two: as a marketer, you used to design the experience on one surface, your website. Now you design it across surfaces, because you adapt to where people want to work on your product. More complexity, but more fun.

Outcome-based pricing

host—JORDAN CHENEVIER-TRUCHET:
You also changed how you price — part of the model is now outcome-based pricing. Everyone talks about it, but it's a big decision: which outcome do you price on? What's the behind-the-scenes?

guest—KIPP BODNAR—CMO, HubSpot:
There's a big theme out there called token maxing — spend as many tokens as possible because it feels like value. It's flawed. The flex should really be outcome maxing: show me massive growth, not quote-unquote efficiency with no growth attached. If you're in the outcome-maxing camp, the question is: what are the right outcomes? It's easier in some places — customer service, did you resolve the question or not; sales, did you book the meeting. It's much harder in marketing, which is why you don't see it there yet. We'll figure it out.

guest—KIPP BODNAR—CMO, HubSpot:
You have to ask: what's the actual value my customer derives from this outcome? Then: I have a wide, diverse set of customers with different business models, so I need to value the outcome in a way that works for everyone and is still economically viable for me. For example, our customer agent is 50 cents per resolution, with a resolution rate in the 70s — a remarkable product priced aggressively, one of the lowest-cost on the market. That's a strategy call: how much do you optimize for short-term revenue versus market share.

Repositioning as a public company

host—JORDAN CHENEVIER-TRUCHET:
You're doing all this repositioning in real time, and you're a public company — investors, Wall Street watching. Usually you'd do this quietly. What do you do differently when every move has to land for both buyers and Wall Street?

guest—KIPP BODNAR—CMO, HubSpot:
I actually think being public is healthy here. You're saying: I want to deliver the value the customer wants and needs, and be a long-term durable solution. The public side just adds time guardrails — there are bad companies that do everything for customers, never monetize, and never make enough progress. So we have an important time horizon and transparency around the company's evolution, and that's okay. I've learned that everything balances out. There will always be periods where certain companies are favored or unfavored. But ultimately, can I provide differentiated value to my customer, and articulate that story clearly? If so, I can get employees, investors and stakeholders on board. The biggest challenge is not doing too much — can you do enough without doing too much?

host—JORDAN CHENEVIER-TRUCHET:
It's not easy, especially with so much FOMO around AI pushing you to do more and more. You want to do more, but you also have to focus on what moves the needle.

What makes a CMO great (from the board)

host—JORDAN CHENEVIER-TRUCHET:
You see CMOs from two angles — running marketing at HubSpot for 15 years, and sitting on boards like Gusto and SimilarWeb. From the board seat, what makes a CMO good versus great?

guest—KIPP BODNAR—CMO, HubSpot:
Being a CMO is a hard job. The first job is realizing the CEO is the CMO — you're in service to the founder and what they're trying to do. Great CMOs understand that but know when to challenge; they don't rubber-stamp everything, and they don't just defer either. They balance their point of view and the founder's into a better joint one. Two: they run their team like Sequoia, not the bond market — an asymmetric portfolio, some bets that fail and some with very high returns, instead of a pile of low-incremental-return bets. The best CMOs go all-in on a few tactics unique to their company that could double, triple or quadruple demand, while still doing the table-stakes stuff.

guest—KIPP BODNAR—CMO, HubSpot:
Great CMOs are also system builders and principle thinkers — closer to engineers than to salespeople. Marketing is much closer to engineering than to sales in mindset. A CMO is supposed to know how to plan a 15,000-person event, build a remarkable website, read API security docs, position products, run a brand campaign, optimize for search, run ads — all of it. If you run a sales team, that's hard, but it's a similar job profile across the team; a CMO needs real breadth. So: can the person learn and get to breadth, and can they hire really well around the topics they don't know, because they'll have blind spots. You can't be great at all of it.

host—JORDAN CHENEVIER-TRUCHET:
That's the main challenge. I tell clients: you think of marketing as one job, but it's fifteen — SEO, paid, and on and on. Nobody is an expert in all of them, which is also why some people can bullshit their way into looking like experts when they're not.

How to lose your board's trust in 2026

host—JORDAN CHENEVIER-TRUCHET:
Last question. What's the best way to lose my board's trust in 2026 if I'm a CMO?

guest—KIPP BODNAR—CMO, HubSpot:
Do what you've always been doing. This is a time for transformation, not status quo. Board members are out in the world doing things differently and talking to others doing things differently. If a CMO comes in and says "this was our plan last year, it's basically the same plan next year, we'll just drag the spreadsheet," that's the surest way to fail — and the most likely, because change is hard and scary, so the easy thing is to keep doing what you're doing. I'd posit it's the surest way to fail.

host—JORDAN CHENEVIER-TRUCHET:
Well, Kipp, thank you very much. For people who want to follow your work — LinkedIn, Marketing Against the Grain.

guest—KIPP BODNAR—CMO, HubSpot:
Marketing Against the Grain on YouTube or wherever you get your podcasts. Hit me up on LinkedIn or X — I'm Kipp Bodnar on both.

host—JORDAN CHENEVIER-TRUCHET:
Thank you, Kipp. Cheers.

guest—KIPP BODNAR—CMO, HubSpot:
Thanks, Jordan. Appreciate you having me. Thanks, everyone.

Introduction

host—JORDAN CHENEVIER-TRUCHET:
Hey Dave, welcome to the show.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
Hey Jordan, thanks for having me.

host—JORDAN CHENEVIER-TRUCHET:
Dave, you're Executive in Residence at Balderton, which closed $1.3 billion in new funds in 2024 — the largest combined early and growth funds ever raised for Europe. Congrats. For people who aren't familiar with the term, it usually means a senior operator embedded inside a fund, advising the portfolio. What's your version of it at Balderton?

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
It asks me to make a distinction between entrepreneur in residence and executive in residence. It's a very fine distinction, lost on most people, but I'm technically an executive in residence. An entrepreneur in residence is, as you said, an entrepreneur — they're looking to get back into the game, they're watching the deal flow and going: ooh, I want to join this team. And for that reason, some people are threatened by them. Is this our new CEO that the VCs are going to inject into my company? At Balderton, we went for the executive in residence model because we didn't want that kind of fear. More importantly, we recruited people who don't want to get back into the game. We enjoy playing the game, but none of us — there are four of us — is looking for a full-time job inside a software company. Been there, done that. Our model is highly experienced people who, for whatever personal reasons, still want to be involved and share their knowledge.

host—JORDAN CHENEVIER-TRUCHET:
Just because I'm curious: how does an American end up advising a European VC?

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
That is unlikely. I think I'm the only Balderton team member who works in California, or even in the US. How does that happen? I know Bernard Liautaud, the managing director, because he was a co-founder of Business Objects. I spent nine years at Business Objects, five of which in Paris — I moved there shortly after the IPO, spent five years, moved back, then four more years working out of our US headquarters in San Jose. So really two things: one, it built my network with people like Bernard. And two, let's just say it gave me special skills about what it's like to build a European software company in the United States.

Ingres vs Oracle: why the best product doesn't win

host—JORDAN CHENEVIER-TRUCHET:
Let's dive into marketing. I think you've got a story to tell about Oracle, back when it was Relational Software at thirty million dollars. What did you see there that pulled you into marketing rather than product?

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
This could be a long story. I was a self-taught programmer — programming on paper tape, over modems, to IBM mainframes, as a child. I paid for college by programming at Lawrence Berkeley Lab. Then I needed a job and ended up at Ingres, one of the original four relational database vendors, then known as Relational Technology. And we had this nasty competitor across the bay called Relational Software. Their product was called Oracle — if you go back far enough, the company was Relational Software and the product was Oracle. Over the next seven years at Ingres, I moved from tech support into competitive analysis into product marketing. But the more important thing was: the company was losing to Oracle, and that was very confusing to me. We were founded by four UC Berkeley professors, one of whom would win the Turing Award 40 years later. The very first CTO I ever worked with is going to be the best of my whole life, bar none. And we kept losing anyway.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
That, to me, was intriguing. I just said: whatever this sales and marketing thing is, it seems to matter a lot, and I'm going to go learn it. Because we had all the technology in the world, yet we were losing — commandingly, not just a little. I went on a sales call once where the customer said: Dave, before you begin, you should know that if you prove to us you're as good as Oracle, you lose. If you convince us you're 20% better than Oracle, you lose. And maybe, if you convince us you're 50% better than Oracle, you've got a chance. The first thing that went through my mind was: gosh, we're going to lose this deal. The second: how could I get my company into that position? Wouldn't it be nice to be the other guy in this transaction?

host—JORDAN CHENEVIER-TRUCHET:
So this is when you realized sales and marketing was everything in the company.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
Yeah. A lot of founders to this day — particularly European founders — would dispute that notion. But I had imprinted on my brain, very early in my career, that best product doesn't win. Therefore there's this thing called sales and marketing, and I'm going to go learn it. For a bunch of personality reasons I'm more suited to marketing than sales, so I built my career in marketing. But that was why: because I watched Ingres lose.

"Hey marketing, go get this" — 30 years of moving targets

host—JORDAN CHENEVIER-TRUCHET:
For people who don't know, you write a lot — you've got a blog, which you recently modernized. In one of your pieces you described the evolution of software marketing as shifting from "go get leads" to "go get customers who renew and expand". If you had to write that chapter today, what's the next "hey marketing, go get this"?

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
I won't get it perfect because I'm doing it from memory. When I started, marketing departments were told: hey marketing, go get leads. You'd go to a trade show, leads were on paper, and the number of boxes was the success metric — hey, we filled two boxes of leads. Then somebody figured out that a lot of those were students, analysts, consultants. They were never going to buy anything. So it became: go get qualified leads. Then CRM came along and it was: go get opportunities. Then somebody said not all opportunities are created equal, we'd like bigger ones — so go get pipeline. Then: not all pipeline is created equal, go get weighted pipeline. Then: weighted pipeline doesn't convert, so go get pipeline from deals that will close. Then closing wasn't good enough — we care about NRR here — so go get opportunities that turn into pipeline, that we win, and that renew and expand on renewal. That's literally 30 years of my marketing career distilled into "hey marketing, go get this".

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
So what are they telling marketing to get now? A lot of this is driven by your pricing model. To the extent people are moving to outcome-based pricing, it's going to be: go get customers who drive outcomes. Customers who are successful using the product, and who use a lot of it over time.

host—JORDAN CHENEVIER-TRUCHET:
What do you think of organizations still asking marketing to go get leads?

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
Part of this is supremely frustrating to me, because any good marketer would know the following. It could be 1989 and I could have a box full of leads — look, you want the team fired up about getting leads, that's how you make a trade show work. You pump everybody up: go run cards, go talk to people. But then you filter them quickly and throw away the bad ones, because as a good CMO, you know the point is sales. I worry a lot of bad marketers forget the point. There's this big MQL backlash now — "the MQL era is dead". Well, MQLs were the only thing you ever wanted if you were a bad marketer. A good marketer knows you want MQLs that turn into deals. A lot of the rhetoric around marketing — the MQL era is dead, performance marketing is dead, it needs to be all brand — is founded on these miscomprehensions. If you always understood that the point of marketing is to make sales easier, you wouldn't get this wrong. So if it's 7:55am and the booth opens at eight: yes, go get leads. But that was never the raison d'être of marketing.

Outcome-based pricing: what is an outcome anyway?

host—JORDAN CHENEVIER-TRUCHET:
Going back to outcome-based pricing — I see two types of companies now. It works really well if you help a company produce a lot: that's what we see with the LLMs, where the outcome is the tokens you burn. On the other side, you've got revenue or results as the outcome, and as a vendor you take a lot of risk attaching your pricing to that. I feel a lot of companies will choose production over revenue as their outcome. What kind of risk are they taking?

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
I love analogies, so here's mine for outcome-based pricing. Say we run a dating site. What's an outcome? Is it a first meeting? A date? A relationship? An engagement? A marriage? A marriage that lasts ten years? Back to "hey marketing" — marriage is not good enough. It's easy to get simplistic about outcomes; even in that simple example, what is the outcome anyway? And by the way, if I know the conversion rates, I can build you a pricing model that's mathematically equivalent at every level. You want to buy dates, they cost this much; you want to buy engagements, they cost a lot more. If I know the funnel, I can price it to be mathematically equivalent on average — it's not hard. So one question is what the customer wants to buy. Most people are optimistic, so they'll want to buy dates, because they're cheaper and they overestimate their conversion ratio. There's a lot of psychology in the pricing.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
Then: some companies are not blessed. Twilio — you send an SMS message. Is that consumption or is that an outcome? Twilio can't guarantee the outcome. If the SMS is a marketing promotion, they're not in the business of getting paid on whether that promotion works. I call these intermediate outcomes. If you're Twilio, it's probably an outcome — we sent the message, we did our job. If you're the marketing software built on top of Twilio, it's not an outcome, it's an enabler. So: one, outcomes are elusive. Two, not everybody has them to price. The best thing in the world is selling resolved cases — that's why everyone uses support as the example for outcome-based pricing. You can do it with a chatbot, and there's an equivalent human cost. SDRs converting MQLs to opportunities: same thing. Those are great use cases because it really is a business outcome, and I know exactly what it would have cost with a human. Not everybody is in that situation — that's why Salesforce created the agentic work unit. Everyone's struggling to find something above tokens, which is really a language-processing metric. But it can't all be closed deals either: if you're writing bad marketing content and I send your message, it's not my fault if it doesn't close. So they're trying to create these intermediate layers.

Marketing exists to make sales easier

host—JORDAN CHENEVIER-TRUCHET:
You mentioned that marketing has to make sales easier, and I know a lot of CMOs who don't really think that way. It's a principle you picked up at Ingres. In 2026, what makes the job of sales easier? More data? More content? More AI-generated pipeline?

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
First, the basics of the principle. I was in a meeting decades ago as a product marketing manager. We'd brought in a new VP from Oracle — Chris Greendale, charismatic guy. He stood up and just said: why do we exist? We exist to make sales easier. Literally, as soon as I heard it, I thought: I like this guy, and I like this mantra. In consumer products marketing, the PMMs are the brand managers and they're the bosses — sales are soldiers they deploy. And a lot of tech marketers deep down wish for that power. So there's this tension: is sales the customer, or is sales our pawns on the chessboard? In an enterprise software company — given how much money salespeople make, how much the department costs, how much power they have — it's a really bad idea to treat them as pawns. You will not last. Go sell yogurt or candy bars if you want sales to be your pawns.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
A couple of years later, I worked at a company where the head of sales came in one day with t-shirts that said: code, sell, or get out of the way. I thought it was fantastic. Two people make money for this company — the people who write the product and the people who sell it — and everybody else is help. Even the CEO is help in that definition, unless they're on a sales call. I later wrote a post called Enterprise Software as a Two-Engine Plane: two engines of flight, code-writing developers and quota-carrying salespeople. That's always been my philosophy. And the first thing about it is that it's egoless: yeah, we're help, and we're proud of it. We're here to help you — which means we need to ask if we're helping, and measure if we're helping. There's a psychological part some people still resist, because they're too proud to admit they're help. If the company had only three people, there'd be a CEO, one developer and one salesperson. Zero marketers. And by the way, that mantra — from the second I heard it — took me from PMM to CMO of a billion-dollar company. It works. I'm a big testimonial for it.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
Now, how do we make sales easier today? Take a funnel view. You can put more in at the top of the funnel, with higher quality — and the CFO will like you if you do it with higher efficiency. Most marketers are very focused on top of funnel, arguably too much. I'd even say they're one level down: focused on what marketing circles call demand capture. You didn't actually generate the demand — you found a guy who was out shopping. That's very important, that's what builds pipeline. But demand generation matters on top of that. You may know Alice de Courcy from Cognism — she did a ton of really good content on demand gen versus demand capture. Everything you want to know about that, go ask Alice. But being a product marketer by background: there's a lot of value you can add lower in the funnel. Better messaging, better competitive messaging, better tactics. If we know the competitor goes to end users with a demo to show ease of use, we counter by going to IT with a security message. That's not a product message, that's a sales play — but marketing can help with that, working closely with enablement. So I'm a full-funnel marketer: if it's anywhere in the funnel, we can help.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
I viewed myself as a doctor: tell me your problem, I've got this medical bag full of medicines, and let me make the diagnosis. Because salespeople aren't great at diagnosis. Eight out of ten salespeople will tell you the marketing problem is "nobody's heard of us". If I could just get more at-bats, I'd get more runs. That's always true — but is it the most efficient way to get more runs? Maybe we need to increase your on-base percentage instead. There are a lot of other levers. Sometimes I call it tough love, sometimes the doctor-patient relationship: I'm here to help you, but I am a professional, I've studied this for 30 years. So how about you don't tell me the answer — you tell me your symptoms, and I'll tell you your disease.

host—JORDAN CHENEVIER-TRUCHET:
It's interesting that you mention ego. I run a service company, and I see a lot of agency owners who dream of having a SaaS business because it looks sexier — they don't like being at the service of others. I find it pretty noble. Saying marketing has to make the life of sales easier doesn't make it a bad job. It's actually great.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
Totally — if you embrace it. Some people don't. It came very naturally to me, but I've met people who think sales are the pawns. In consumer they are the pawns — I have friends in CPG marketing, I understand where it comes from. If you went to a CPG-oriented business school, you got shaped with the wrong notion of what sales is for enterprise software. But there's nothing better than being help. When you embrace it, you do better at it. How can I help? And then: wait a minute, I'm also a professional. So let's hear your challenges — and I'll apply 20 years of experience to answering that.

AI-native or zombie

host—JORDAN CHENEVIER-TRUCHET:
You wrote that in 2025 most of the VC capital went to AI-native companies, and that traditional SaaS risks becoming zombies. For a CMO working in a traditional SaaS, that sounds risky. What does he or she do in the next few months?

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
Let me explain the point of that post. Back in the old days, you could bluff your future story. If somebody said "you're an on-premises vendor", you could say "but we're building a cloud offering" — and that was good enough. You could show up as a regular on-premises vendor, sell a cloud story, and raise money. People were a lot more patient then. The main point of the post is: that doesn't work anymore. "We're a traditional SaaS vendor but we've got a great AI story under construction" — doesn't work. If you want to position a great AI story, you need a great AI product and a great AI growth rate to go with it. And if you don't, then you should be profitable — a good financial investment. The zombie is the company in the middle: $20 million, growing 13%, break-even. Nothing exciting as a financial investment, an AI story but no AI product — so the story value gets discounted to zero. Without this framing, you could end up there thinking "we got away with it last time, we can sell the story". The answer is no. There's no middle ground anymore. If you're the CMO, all you can do is talk to the CEO: hey, what are we doing? Either we push profitability, or we do something to get an AI product — acquire something, do a tuck-in. But we can't kid ourselves that a deck about an AI product coming in two years is going to impress anybody.

host—JORDAN CHENEVIER-TRUCHET:
Maybe a tricky question. I feel it's really hard right now to tell the difference between a true AI-native product and a wrapper. A lot of companies invest in AI internally or in their product and feel they're AI-native. What's your true definition?

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
A couple of things on what I see out there. The most frustrating place is the people who were into AI about five years ago — call them AI 1.0. They actually had a native AI story — it was just native AI machine learning. The wrong native AI. Those people I feel for; they're deeply confused: "but we are native AI, we got our master's degrees in AI and founded the company". That's the hardest case. Next to it, the traditional SaaS company: just having a copilot or a little bot on the side is not going to be enough. Orchestration might be — it doesn't have to be a thick orchestration layer, as long as it has domain knowledge in it. I believe small SaaS companies do well by knowing a problem space really well. I don't care how thick the technology is. I care how easily someone else could build it — particularly if they don't know the space as well as we do. That's where competitive advantage comes from: we know everything about dentists, we built a relatively thin AI orchestration layer, but it does everything a dentist wants.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
And this is why I want the growth rate — I'm not fundamentally a technologist anymore. I can get into a long theoretical discussion about whether something is AI-native, whether it's thick or thin, what the moat is. But the first test is: is it selling like hotcakes? If it's not selling like hotcakes, I don't care. That's why I say you need a native AI product and a native AI growth rate — and I'm much more flexible on the definition of the product than on the growth rate. Otherwise you're bolting some AI onto the side of an existing offering. That's not bad — it'll probably help you sell your existing offering to existing customers — but you haven't changed anything about the company. If you're trying to say "we've done an AI pivot", I need to see more. And the proof isn't found in the slides. It's found in the growth rate. We could do it with slides before; it doesn't work anymore. People go: great slides, what's the growth rate? Thirty percent? You're not going to be able to raise money.

Trust is the last moat

host—JORDAN CHENEVIER-TRUCHET:
Let's talk about trust. Every year you write predictions, and in your latest ones you said: only trust will get emails opened, podcasts listened to, reviews believed. At the same time we've got AI content everywhere — fake reviews and everything. If trust is the last marketing moat, how do you actually build it?

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
It's a tricky question. I'm somewhat contrarian in general, and I watch the world generating AI slop left and right. There was this era of gated content — I never liked it. My blog has always been ungated, and even with companies I worked with, I was always asking: do we really have to gate this? The better a piece of content is, the more I want it out — I want people to hear our point of view. I would never gate a white paper with our industry point of view. I might gate a benchmark study with numbers people want. Marketing was gate-slash-MQL oriented for so many years — the gate is what captures the MQL — and that's what people mean by "we had it all wrong with MQLs". You didn't have to get it wrong. Just ask intelligently: what's more important here, getting the word out or getting the MQL? If it's a tactical piece and I need MQLs — boom, gate it. But if it's our worldview and why you should move to our kind of architecture, that should never be gated.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
Against that backdrop, you now have the ability to generate a hundred times more content — a hundred times more MQLs! But people don't have a hundred times more room in their brains. And people are very good at detecting AI slop. My primary slop detector is when my brain stops while I'm reading something. What just happened? Why did I fall asleep in the middle of this piece? Then I look at it and see the em-dash, the "it's not this, it's that", all the telltales of AI writing. I've always believed a couple of first principles. One: people are smart, and we should respect that. Two: interested buyers will read long content. This goes all the way back to David Ogilvy — long copy sells. Don't only produce morsels of content because "people don't have time anymore". People have time. If they're betting their job on buying a financial planning system, trust me, they have time. How long is the average episode of Acquired — four hours? People have time, when you've got good content on something they care about. I've always preferred those first principles to "make your blog post 217 words and include these keywords". That stuff doesn't work. AI just put an amplifier on all of this: it's easier than ever to generate ten times more slop. I think it's an opportunity to not be in the slop. Or at least to keep a special reserve — the founder's reserve, the wine cellar: this is the good stuff, ungated, with our real principles. Maybe you do syndicate crap everywhere just to get the leads — I'm a pragmatist, maybe you have to. But somehow you indicate: this is our primary resource library.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
And it means don't make empty marketing claims. How do you lose trust? The expression is: trust is gained in drops and lost in buckets. You gain it very slowly by always being credible, never exaggerating your claims, putting a hard filter on what you produce: does this read like marketing jargon junk, or like something an intelligent person would say? In every little thing you do, you keep the trust level high and you resist the seduction. Because it's lost in buckets — you screw up once and you can lose a year of work. I don't have a simple solution, but: run high-trust communications, and you'll get rewarded over time. Your competitors are going to slop people to death, people will stop opening their emails — and people can tell quality content.

PR is the new SEO

host—JORDAN CHENEVIER-TRUCHET:
You also said PR is the new SEO. Is that the continuation of this thought?

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
That one was more tactical than intended — these are annual predictions, and I think that one lasted about a year. I've always considered SEO trench warfare: it changes every day, you gain a yard, you lose a yard. And AEO is the same. At the time I wrote it, one of the best ways to get good AEO was to be featured in a Wall Street Journal article, a Forbes article, Fortune, the Financial Times. Authoritative big press was one of the best things you could do for AEO. And I just thought it was hilarious — we've gone full circle. For a decade we killed PR, didn't care about PR, and all of a sudden PR is one of the most effective ways to get good AEO. I found it ironic and funny. Now I think that's passed a little. For a while it was Reddit — you had to have your Reddit content. Fact-structured content is important. And I think YouTube is now super important — they're changing their algorithms. So not a multi-year prediction, but hopefully correct for the year I made it.

host—JORDAN CHENEVIER-TRUCHET:
One thing that's hard for most CMOs: PR is this black box with murky ROI. How do you manage it?

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
There's definitely bias in anyone's writing — I think companies underinvest in PR. So not only did I enjoy the irony, it was another angle to say: you should invest in PR. I know two-hundred-million-dollar companies that don't have a PR firm on retainer or a PR person. If you're ten or twenty million, okay. But if you're one hundred, two hundred, five hundred million — I don't know how you don't have a PR team. You have to be careful how you aim it, and PR has changed a lot, but I've always been a huge believer, because it's earned media. It's kind of free: if you're clever and know how to work the system, you can get a lot of awareness and impressions at relatively low cost. It's also good for relationships. The vast majority of market-leading companies are good at PR — it's part of what makes them. You can argue Andreessen Horowitz made themselves with PR: they hired the woman from OutCast to run communications when they were relatively new. What's the cart and what's the horse at Andreessen — the investments or the PR? I think the PR did an enormous job for them. But you're right: it's expensive, it's a black box, you can waste a lot of money. So do it with a contractor, part-time. When your story is good and topical, you can execute — and you're not carrying a fifteen-thousand-dollar-a-month retainer when it's not. There are very experienced PR pros who've retired and still want to work. There's a way to have PR without saying no to PR.

Metrics reflect strategy, they don't drive it

host—JORDAN CHENEVIER-TRUCHET:
There's also the attribution topic — PR is really hard to measure. On this you said: metrics reflect strategy, they don't drive it. I find this claim really interesting, because most boards ask their CMOs to be metrics-driven. How do you manage that tension?

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
If I'm anything, I'm a strategy-first business person. Strategy is a big fancy word, but the best definition I've heard comes from the preface of Rumelt's book, where a professor who's been attending his class says: "it just seems like we spend a lot of time here trying to figure out what's going on." Yes — that's strategy. If you can figure out what's going on, you can know what to do about it. The knowing-what-to-do part is actually not that difficult; the figuring-out-what's-going-on part is exceptionally hard. So we figure out what's going on, build a plan for executing within that, and use metrics to measure progress on the plan.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
To take it back to marketing: we've been through an era very focused on performance marketing, so all the performance marketing metrics are in vogue. If you believe, as I do, that the pendulum is now swinging back towards brand — more and more marketers are saying performance channels are saturated. Google has optimized paid advertising, from their perspective, to the point where it's barely cost-effective for me. Which in some sense is their job: that's the equilibrium they want. The strategic view says: these channels are saturated, we allocate less money to them, more higher in the funnel, more to demand generation than demand capture. And therefore we can't run on the same metrics. The new metric might be market research on awareness, positive opinion, consideration — as opposed to just counting MQLs and opportunities. Marketing is always a triangulation problem; you should look at all of it. But you don't start with "I've got a compass that measures direction" when what we care about is altitude. Wrong tool. Figure out what you want, then find the device that measures it.

host—JORDAN CHENEVIER-TRUCHET:
It's funny — I asked about metrics and you said you're more into strategy. But everything you write is really scientific and mathematical. There are a lot of numbers and calculations.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
Really good observation. It's also true of execution. People say: Dave, you call yourself a strategist, but you're an execution fanatic and a metrics fanatic — help me reconcile this. The answer is pretty simple. Execution first: if we execute badly and something fails, I learn nothing. I don't know if it was a good idea or not. I always say: you're spitting in my petri dish. Bad execution ruins the experiment. I'm trying to figure out if we can sell this thing to mid-sized insurance companies, I've got three sales reps trying it for six to twelve months — and if you hire three clowns who don't show up to meetings and we don't sell anything, I learned nothing. You let your bad execution blow up my experiment, and now I've truly wasted a few million dollars. Whereas if we hired three aces, they did their best, and it didn't work: okay, we can safely exclude mid-market insurance from our target markets. I feel the same way about metrics: they let me measure progress in the experiment. They're my way of knowing, objectively, whether something's working. And a lot of my favorite measurements are actually strategic — market research: have people heard of us? What do they think about us? Why aren't they buying us?

Three ways to get fired as a CMO

host—JORDAN CHENEVIER-TRUCHET:
You wrote about the three ways to get fired as a CEO. I wrote them down:

• "I'm getting tired of running the company."
• "I'm running out of ideas for how to fix our core problems."
• "I think we need to sell the company."

host—JORDAN CHENEVIER-TRUCHET:
What's the equivalent for CMOs?

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
We were talking about this earlier — I think we came up with a good list. One: "I don't want to be accountable for a pipeline target." Great way to get fired. Two: "I don't want to do attribution, because it's too hard and it doesn't really work." That one's more subtle, because attribution is imperfect. But as I always say: until the CEO, the CFO and the board stop asking, you need to keep answering. If you know a better way to answer, please tell me what it is. I'll be the first to concede that attribution is imperfect — but absolutely necessary. You need a kind of schizo view on it. The business will not stop asking, and it's not good enough to say "it's really hard to measure". Some CMOs try — it's a good way to get fired. It's a weird learning cycle: when you know a little about attribution, you're super jacked because you think it works. Then you learn more, discover all the limitations, and go super negative. Then you need to come back up: people need it anyway, I need to provide it anyway, and my job is to limit the misinterpretation of attribution data when I present it. That's the actual challenge of the CMO. And third: "there's nothing I can do about win rate." A lot of marketers are too top-of-funnel oriented. If we're only winning 5 or 10% of our stage-two opportunities, that's not good enough. I'm Doctor Marketing, I'm here with my bag — let's look at what's going wrong and see if we can get that win rate up.

host—JORDAN CHENEVIER-TRUCHET:
What's interesting is that we can sum them up as: they don't take accountability, or they don't realize the impact they can have on sales.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
True — refusal to take accountability covers the first two. And on win rate, I overstated it: it's both a marketing and a sales problem. If we can help, it's part of our problem.

The dismantling of the CMO role

host—JORDAN CHENEVIER-TRUCHET:
You've written about the decomposition of marketing into distinct pillars: PMM joining product, demand gen joining sales, brand joining the CEO. What's your view of the marketing department in 2030? Because if we go this way, there is no marketing department.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
Just to be clear: I wasn't recommending this as the ideal marketing structure. I was saying — and this is true — I'm seeing it more. Literally, I had one CEO say to me: Dave, I've had four CMOs in five years. I can't do it again. Don't make me touch the hot stove again. I don't want to hire a fifth CMO. That's the only case where I've actually recommended it. But I've seen it happen a lot. You put product marketing under product — that works when you have a marketing-y product head. We did it at Salesforce: all the product GMs, and I was one of them, were pretty marketing-oriented people. It creates a problem: if you have multiple products under four different GMs, who tells the whole story? That creates a need for what I call product-line marketing in the central org. You can put demand gen under sales — I don't love it, because most salespeople aren't that interested in it, but you can do it.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
My test for when it works: if the CRO says "Dave, I want to be president and then CEO one day. I want to learn marketing, and integrate it better into my org" — I love it. That's a good reason. If it's just self-preservation — "those guys can't generate enough pipeline for me, and my health was at risk, so I'm going to own it" — I don't like it. We can find a CMO who will care about you. If the problem was a CMO who didn't care about demand gen, fix it by replacing the CMO, not by moving demand gen under the CRO. And brand and comms typically end up under the CEO. In those orgs you have no CMO — or you call the brand-and-comms head "CMO", the hollowed-out CMO job, which is not uncommon. Look, I don't recommend this except in "I can't do it again" situations. But I'm seeing it happen more and more, and I understand why: CEOs are frustrated they can't find the all-sport-athlete CMO. They can't find it, so they fail, and they decompose the role. I hope the pendulum comes back on this one — I believe in CMOs. But right now, this is what's happening in the industry.

What European founders get wrong about the US

host—JORDAN CHENEVIER-TRUCHET:
Let's take a step back. You've worked both sides of the Atlantic. What's the most misunderstood difference a European founder has about the US market?

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
I'll give you two. One: the commitment required, in time, money and effort. They underestimate how big a deal it is — including Business Objects, where I did it firsthand. Two: the importance of marketing, image, brand. If I have to grossly stereotype product-oriented European founders: they have yet to learn the lesson I learned at Ingres — best product doesn't win. They're smart enough now not to say it. Ask them "do you think the best product wins?" and they'll say no, other things matter. But deep down, I think they still believe it. Deep down, American CEOs understand that marketing and sales have a huge impact on the outcome. European founders might even be in a more dangerous middle ground: they accept it verbally, or pretend to understand it, but they don't. They still think that if we show up with a good product and show it to people, we'll win. That's simply not true — I don't think it's true in Europe for that matter, and it's definitely not true in the US.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
And I have to mention one more, because it comes up so much: the value of their references. Unless you're a vertical player whose European customers are known worldwide in the industry — if you're a horizontal player and you show up saying "we've got great references: Bouygues, Carrefour, ASML, Reed Elsevier" — not only have most Americans not heard of these companies, they can't pronounce them. These references are just not useful. And that's really painful for the founders, because they built these amazing accounts into big successes, and outside their vertical they're worth nothing. So try to get global brands on your European customer base, or accept that you're starting from scratch in the US. Schneider Electric is not a reference that's going to carry you real far over there.

Start in New York, relocate a co-founder

host—JORDAN CHENEVIER-TRUCHET:
Say I'm a European scale-up founder and I want to go to the US. Where do I start? I understand I need my main use case in the US — go shake hands, get my first customer, make it a really good use case, then take it to other companies in the same industry. Correct?

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
Yeah — that's Geoffrey Moore 101, the bowling alley strategy. Find a use case, find an industry, and go press the flesh. But let me give you an alternative tactical view: where do you start? New York. For eight out of ten European startups, New York is the answer.

host—JORDAN CHENEVIER-TRUCHET:
Why? Especially in tech, I'd think San Francisco.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
What's the New York advantage over San Francisco? Closer flights, less time difference, the most culturally compatible city. San Francisco is pretty culturally compatible too. But you want to save money and open in Atlanta or North Carolina or the Midwest? I just think it's harder for European founders and their partners to be happy living in Atlanta compared to New York. And those things matter, because if we're going to relocate a founder — which I also recommend — they have to stay. They can't leave after six months because their partner says "I've got to get out of here". New York is a good location, you can find talent, and there are multiple industries all around: pharma in New Jersey, insurance in Connecticut, finance in New York. Great flights across the US. It's just a really good hub. At Business Objects I flew Paris–San Francisco almost once a month for nine years, and I can tell you: it's tiring, and the time difference is bad. Now, if I'm a hot AI company, I may have to do San Francisco — that's the only exception. The AI scene is happening there, so if you're related to foundation models or core AI technology, that trumps New York. But for a typical company: open a New York office, and move a co-founder there.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
And then I've got to talk to you about how to read US resumes. Probably the single funniest thing a European founder ever said to me — he was looking at a pile of American resumes — was: "these people are gods." I said: no, that's just a schleppy sales rep from New Jersey. You think they're gods because you don't know how to read American resumes. American resumes are embellished. Europeans understate their accomplishments; Americans overstate them. So you look at an American resume with a European eye and think: these people are gods. No, they're not — and we need to teach you how to interview them. Behavioral interviewing is hugely important here, because it forces people to tell stories about actual things, and it cuts through the embellishment. If you ever wondered why behavioral interviewing is so popular in the US: that's why. US resumes are super inflated, and this is the antidote.

host—JORDAN CHENEVIER-TRUCHET:
It's funny you say this — all the French founders I know who went to the US to hire salespeople told me exactly this. A hundred percent of the time.

What Dave looks for in founders

host—JORDAN CHENEVIER-TRUCHET:
One last question. What's the one thing you look for in deciding to work with a founder — or giving a positive rating in a diligence process?

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
That one's tricky. My first rule of founders: all founders are crazy. They have to be, definitionally. They're quitting a perfectly reasonable job to start a company that's a one-in-a-hundred or one-in-a-thousand shot. Sometimes VCs wish for sane founders — that's just not going to happen, they don't make them that way. So you need to pick what kind of crazy you want. They come in different flavors of crazy, and I like some flavors better than others. That's the first principle — not a qualifying criterion, an ambient criterion. If I had to pick one thing I like: curiosity. I'm a huge believer that curiosity is the number one thing in sales. And to the extent founders need to sell, curiosity is just so powerful. You want to learn about what you don't know — that makes you a better founder, because you're always learning, you're not pretending to anybody that you know everything. "Hey, I'm a product guy, I built this thing. I know a ton about the product, a little about the space. I'm curious — how would you use it?" Almost every single important question in a sales cycle, you can get to just by saying "I'm curious": why do you want to buy this? Why did you call us? Why would you ever buy from a little French startup? Have you ever bought one of these before? Has anything changed? It also applies to becoming a better CEO — going to your investors and mentors: I'm curious, how have other CEOs handled this situation? And being curious with customers about their problems: why is this a problem? How have you tried to solve it before? When you're done solving it, what's the next thing you do? That general curiosity helps you find where you fit in the value chain, and it helps you find adjacent expansion opportunities. So if there's one thing: always be curious. What don't I like? Defensive founders. Proud founders — people too proud to admit they don't know what they don't know. They're going to be too proud in sales calls too, they won't find out, and they're going to lose. So curiosity is the thing.

host—JORDAN CHENEVIER-TRUCHET:
Thank you very much, Dave, it was great. For people who want to know more about what you're doing or contact you, we send them to your blog and your LinkedIn.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
Yeah — kellblog.com, my LinkedIn at /kelloggdave, or my Balderton email if you want to reach me: dkellogg@balderton.com.

host—JORDAN CHENEVIER-TRUCHET:
We'll put all the links and contacts in the description of the episode. Thank you again, Dave. See you next time.

guest—DAVE KELLOGG—Executive in Residence @ Balderton Capital:
Thanks, Jordan. Thanks for having me.

Introduction

host—JORDAN CHENEVIER-TRUCHET:
Hey everyone, and welcome to the show.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
Thanks for having me.

host—JORDAN CHENEVIER-TRUCHET:
Bruno, you've been at Clay about two and a half years now. Since you arrived, everybody I speak to in B2B and every company I work with has heard about Clay — they want to work with Clay, they're Clay users. You've managed to be everywhere and make sure people talk about you. You've got this content playbook: founders speaking online, the team, creators, agencies. How do you manage the whole thing?

Two engines behind Clay's content machine

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
There are two sides of this. One is the content that people in our community create — or people in general who aren't even associated with Clay. The other is the content that we create ourselves.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
Let me start with the community side, because that's the one people have the most questions about: how are we getting all these people to talk about Clay? The first element is that Clay is a very flexible and powerful product. When you look at the market for sales automation and data enrichment in B2B before 2023, most products were inflexible, with a sales-led motion. Clay came in self-serve, very flexible and horizontal. So people could actually build on top of it. The product unlocked a lot of creativity — people could create go-to-market workflows they could never build before.

Why the ecosystem posts: brand association over cash

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
That's the foundational layer. As people build these things, a portion of them are service providers — they have freelance businesses, agencies, or they're individual contributors who want career progression. Clay enabled them to showcase to the market that they were very innovative. I saw a similar pattern before Clay, during my five years at Webflow, and on design and developer tools too.

host—JORDAN CHENEVIER-TRUCHET:
Notion is a good example of that.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
Notion, Airtable early days, Webflow, Framer, Figma. The flexibility lets people showcase. The tricky part as a marketer is: how do you have any control? The way we found is by creating programs in the community that incentivize people in a soft way. And I say soft because — this was a learning — most of these people don't want “post this and I'll give you X.” What they want is brand association with us.

host—JORDAN CHENEVIER-TRUCHET:
So not basic sponsoring.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
Right. Clay was innovative in the market and people wanted to be associated with that. So the question became: can we build programs that give this brand association, which then increases the output of content they create? We built the Experts program — now the Solutions Partner program — the Creators program for people getting started, and a lot of educational programs. The education piece is the core. They learn Clay, understand the value, get invested in the brand, create content, and distribute it on a channel.

Authenticity, and mapping employees to personas

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
Then there's the content we create — our executives, our employees. What we noticed early is that AI is unlocking content creation like never before. Everyone can create content today. The downside is no one is authentic. So people, more than ever, want to hear from people, and they want content that's clearly not AI.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
Every company can do what we did. Clay has two sides of the business: self-serve and sales-led. Self-serve has a lot of startup founders, agency owners, and first marketers at a startup. Sales-led typically has rev ops and marketing ops as the end user, and CROs and CMOs as the final decision-makers. So we selected specific individuals in our company that match those profiles, and put real resources behind them to create content on LinkedIn relevant to that audience.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
For the ops persona, we have Everett, our head of go-to-market engineering. For the founder persona, Varun. For the marketing persona, me. For the CRO persona, Karan, who's our CFO and CRO-ish. And for SDRs, our head of SDR. We pair actual writers with these people, meeting weekly — “here's an experience I had, here's what I want to write about” — and they go back and forth and post.

Everything funnels into LinkedIn

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
The main thing is that everything funnels into LinkedIn, because we keep seeing it's the channel where buyers spend time — we hear it constantly in how customers find us. That's why people have the impression: holy shit, I'm seeing Clay and everyone's talking about them.

Positioning a flexible, unopinionated product

host—JORDAN CHENEVIER-TRUCHET:
I had a conversation with your team, and Nancy told me you don't want to be opinionated as a product. That's interesting — Figma and Notion have to be unopinionated to offer flexibility. But when you do GTM for a team, you kind of know the playbooks. Still, you decided not to be opinionated. How do you play with that in marketing? People expect you to be opinionated because you do GTM.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
Having a horizontal, flexible product is the difficult part from a marketing perspective. The difficulty is positioning: the product is so flexible that people build insane use cases. Some use Clay as the backend of tools they generate. Some use it for outbound, some for inbound. So how do you position it? That's the challenge — and also the exciting part, because Clay can serve specific problem areas really well. We can verticalize it to different industries. If you're targeting a dentist office, you can get data points you'd never get otherwise, because of the flexibility.

The lifecycle of every tactic

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
We're not opinionated on how you build, and I think that's where a lot of GTM tools failed. The ones that took off — like Salesforce — are flexible. Every business is different. The data point you need to identify your ICP with 100% confidence is often very specific, and you need a flexible product to find it. Most other go-to-market products were built around one playbook that worked for one type of company — usually tied to a channel.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
But there's always a lifecycle to a tactic. It starts effective, then everyone adopts it, and we destroy the effectiveness of that channel. So you need a tool that lets you keep pushing the boundaries. I bought this, it worked for a while, now it doesn't — what else can I do? The best companies at go-to-market are the ones pushing and trying new things as frequently as possible.

From “hacky tool” to legit

host—JORDAN CHENEVIER-TRUCHET:
When you say companies built around one use case that then disappeared, it makes me think of the whole growth-hacking wave. Ten years ago on LinkedIn you'd take your cookie, drop it into a tool to scrape people. All those companies died, because now it's so easy.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
That was a huge thing when I joined Clay at the end of 2023. A lot of people were associating us with those tools — “is this even legal? It's hacky.” A big focus for us was elevating the product to show: we're not in that group, we're a legit product that legit companies use. We pushed the brand to stay a little weird, but less hacky — this is an actual product and tool.

host—JORDAN CHENEVIER-TRUCHET:
And there was always that techie, geeky vibe. With Clay everything feels smooth — you don't need to be a cybersecurity expert to use it.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
Exactly. With that old wave it was always “oh my god, can you actually do this?” — and six months later it all got shut down.

A nine-year overnight success

host—JORDAN CHENEVIER-TRUCHET:
A few months ago Varun, your CEO, posted about why the company took so long to get to $1M ARR and then went explosive. I commented that at some point you really nailed your positioning. For a few years I was like, what's the point of this product? Then everything clicked.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
For those who don't know, Clay has been around since 2017. It took about nine years to get to $1M ARR — and then from two years ago, $1M to $100M. When we announced $100M ARR, I called it a nine-year overnight success. Everyone says “we're the fastest growing.” Sure, we're fast in the last two years. But there's a backstory of seven years where we were almost dying.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
Clay was this super flexible product — a spreadsheet that built websites, no-code. But we didn't have focus; we were spreading ourselves too thin. Early 2022, when Varun came in, he understood we should focus on the go-to-market use case. A lot of people were already using Clay for data enrichment and go-to-market workflows. In 2023 we focused there, and I joined later that year to scale it. Lack of focus is the death of a lot of early-stage startups. You have to focus on a core use case.

Inventing the “GTM Engineer”

host—JORDAN CHENEVIER-TRUCHET:
When growth hacking was a thing, there was the growth hacker. Now with Clay there's the GTM engineer. You literally helped invent a job title. Tell me the story.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
The rules are evolving fast with AI and tools like Clay. Before this wave, you had a growth or performance marketer who owned specific channels — Google Ads, SEO, CRO, lifecycle. And you had ops people who controlled the data pipelines and the CRM. What we saw early, at companies like Ramp and Rippling, is that the best people doing interesting go-to-market work were a mix of both. They were closer to ops than to a typical channel marketer — more technical, understood APIs, webhooks, all the complexity of sending data from one place to another and launching programmatic motions.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
So there was a skill gap to fill to operate at high pace. Every company defined it differently, but they all had this person. We called it the GTM engineer: the person who builds systems and automation that generate revenue — top of funnel, conversion, or expansion. Then we started creating content: here's what we're seeing. The interesting part is you need bottoms-up adoption. You tell the market what you're seeing, then watch: are people calling themselves this? Do they resonate with it? If they do, it's real. If they don't, it's just a marketing campaign. People resonated, called themselves this, companies realized there was a gap, and they opened roles.

Defending the role

host—JORDAN CHENEVIER-TRUCHET:
Tough question: for some people the GTM engineering role was built and marketed because it fits perfectly with your product. For some companies it's not obvious the role is needed. How do you defend against that?

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
Some companies call it growth engineering, some call it business systems. You can call it whatever you want — we call it GTM engineering because it's closest to the problem these people solve. I hear “isn't that just rev ops, or what growth marketing should have been?” Sure, maybe it should. But is it being done? If not, you need someone to do this work.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
I don't think it's one or the other. GTM engineering often sits under ops, but there's still a lot of scope maintaining systems, keeping data clean, keeping the stack running. You also need someone thinking about systems that generate revenue and enabling the team to build them. The problem statement has to be there: you have to build systems that integrate all your channels to generate revenue — especially as top-of-funnel channels become more automated and you need fluency connecting them.

Naming roles to attract the right people

host—JORDAN CHENEVIER-TRUCHET:
I think marketing a job title is a good thing. A lot of people call themselves a CMO because they want to be the top marketing person, but they don't really know what a CMO does. You can call it growth marketing, rev ops, or GTM engineering — but are we talking about the same thing? If you market a job description, will you attract the right people? Organizing around one job title helps everyone agree on the same tasks and objectives.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
100%. If you call it the wrong way, you might not attract the right person.

host—JORDAN CHENEVIER-TRUCHET:
For my own company right now, we're recruiting. We know the role — some call it head of growth, some head of marketing. So we use the same job description with different job titles, because the people we want to attract describe themselves differently.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
Smart. That's cool.

Controlling the narrative — and measuring it

host—JORDAN CHENEVIER-TRUCHET:
You identify creators, agencies, and you push internally for people to post. But do you push different narratives, double-check what gets posted? What control do you have over what people post about Clay?

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
Internally, we have someone who oversees executive content — we just hired for it. Mapping people to a specific persona gives them guardrails and direction; that's basically the only direction we give, plus it needs to be something genuinely interesting that adds value. For the external community, we don't really control it. We share our positioning — here's how we think about things — but it's weird for me to tell people what to post. It's their personality, their channel. Burn that bridge and it's over.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
Some people position Clay a little off, and I'm okay with that, as long as we have people talking about us positively. Here's the key thing: you can have the right channel and people talking about you, but those two alone aren't enough. You need high frequency. If it all happens in one week at the start of the year, three months later everyone forgot about you. So you need people talking about you all the time. It's our job to give top-down direction as much as possible, but you can't fully control it — you keep the momentum going.

host—JORDAN CHENEVIER-TRUCHET:
And how do you measure this?

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
We measure top of funnel through what we call qualified engagements on LinkedIn. A post has to mention Clay. Then we look at the volume of likes and engagement on those posts. We have a named-account list of who we want to go after, and we cross-reference it with the people engaging — are we attracting high-quality people? Early on we measured by raw social engagement, and some posts got tons of engagement from weird bots. That was the wrong thing to do.

host—JORDAN CHENEVIER-TRUCHET:
It's clickbait, and the engagement is bad.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
It shifted us to invest more in executive content that fits the personas, because that attracts really high-quality people. That's how we measure the top of funnel.

Brute force vs. analytical sophistication

host—JORDAN CHENEVIER-TRUCHET:
In your Understory interview, you said importing Webflow's complex growth processes to Clay too early was a mistake — early stage needs brute-force execution over analytical sophistication. But a lot of listeners are CMOs and heads of marketing past early stage. What's the actual breakpoint?

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
At Webflow I led the growth team owning self-serve acquisition — a big self-serve business. I had bi-weekly sprint planning with engineering and data science, weekly metrics reviews with leadership, a whole prioritization motion. Looking back, it made us a little stiff. When I joined Clay, I knew the playbook, so I set those things up. But the speed was so different, and Clay was 18 people — my team at Webflow was 18. Things moved so fast, at a different scale, that everything behaved differently. So I realized this doesn't work here. What mattered was shipping things — create the programs, build the motion — and then we can measure.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
For context, Clay is now around 400 people, with a marketing team of about 30. We're later stage, so we do look at the numbers — weekly metrics, a monthly retro with founders on the operating cadence. But my obsession with the marketing team is: how do we not lose speed as we grow? It's easy as you scale to become obsessed with attribution, locked into a motion where you have a meeting to present, then a meeting to prep the meeting, spending all your time pulling data.

Conviction over attribution

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
Do you have conviction on the channels and bets you're making to solve a specific problem? We have four bets for the next six months: grow the product-led motion, grow the fast-growing sales-led motion, verticalize Clay into new industries, and international expansion. Those are the problems to solve. Marketing-wise, do you have bets you're convinced will help solve each one?

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
For international expansion into Europe, we don't want a top-down motion — we want to invest in community and bottoms-up events. How do you quantify that? It's the right thing from intuition and a business perspective; we have conviction. But quantifying the revenue impact is really hard. So you need conviction on the bets and problem areas to execute, while reviewing pacing bi-weekly or monthly. Some areas — the growth side, pipeline — are much easier to quantify, and there you should be focused on hitting the numbers.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
But people get bogged down there and forget how buyers actually decide to buy software. You know Clay because you saw a lot of people talking about us — not because of some programmatic channel. So what are the inputs that help us be everywhere within the right audience? Those are hard to track, but you have to have conviction. Luckily, our founders really understand the value of brand and content, and how it's different when it comes to attribution. That's the hard part of the CMO–founder relationship: this is the right thing to do, here's why, it's qualitative — are we aligned?

host—JORDAN CHENEVIER-TRUCHET:
It makes me think of Dave Gerhardt — you need to work for a CEO who gets marketing.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
Exactly.

The seven-figure billboard bet

host—JORDAN CHENEVIER-TRUCHET:
Since we're on attribution: last year you dropped seven figures on physical billboards. What did the decision process look like, and what was the leading indicator that told you it worked?

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
What's maybe different about us on billboards is we don't run them all year. We do a specific window in a specific city where we know there's a strong concentration of our core ICP. The window is late August to early November. A lot of our ICP is in San Francisco then — Dreamforce, SaaStr, the a16z conference, TechCrunch. Everything happens in that timeframe. So how can we add a touch point for people who are already seeing us on LinkedIn?

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
In 2024 we did our first billboard campaign, low six figures, same strategy. The problem we needed to solve was the same: how do we show Clay isn't a hacky product? There are channels like billboards where, no matter how big you are, people see you and think “these guys are legit.” You seem like a big company even if you're not.

host—JORDAN CHENEVIER-TRUCHET:
It seems serious.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
So we increased the spend to keep elevating the brand. There are tactics to increase the likelihood of reaching the right person: we mapped all our named accounts in the region — where are their offices — and got the best billboards near them, supplemented with events at the same time, and sponsored those conferences. Everyone who flies in sees us at the airport, on the highway, downtown, at the conference, at our event.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
Tracking this is the hardest thing ever. The best way we found is through Gong calls — open opportunities where they mention they saw us there — and we attach the pipeline generated to that campaign. We tried lift studies, but it's really difficult. There's also a secondary benefit: billboards help a lot with recruiting. People see it and think, this company is cool, I want to work there.

Marketing is what your actions suggest

host—JORDAN CHENEVIER-TRUCHET:
Our job in marketing is to think about what our actions suggest about us. By doing billboards, you suggest you're big and serious. I always give this example: I'm the co-author of a marketing book. Most marketers haven't read it, but the fact I wrote it means, to some people, I'm a marketing specialist — even if I were a bad marketer.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
It gives you authority.

What's next for Clay

host—JORDAN CHENEVIER-TRUCHET:
Last questions. What's your focus for the next few months or years? What's the next big thing?

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
Two things. First, international expansion. Clay naturally has a lot of pull, especially in Europe, with a lot of self-serve customers. How do we build a marketing motion that isn't tone-deaf and resonates with people in different markets — and then as we move into APAC? Second, how do we grow beyond tech? Today Clay is well known inside our bubble — most B2B tech companies know us. But the most legit companies go past tech. Is a random manufacturing company somewhere considering Clay? Getting into those industries is a core question. Those two bets give us step-change growth.

host—JORDAN CHENEVIER-TRUCHET:
Good luck with that — it sounds exciting. Thank you, Bruno. For people who want to get in touch, should we send them to your LinkedIn?

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
LinkedIn is the best place.

host—JORDAN CHENEVIER-TRUCHET:
Thank you again, Bruno, for everything. Keep in touch, and good luck with your next challenges.

guest—BRUNO ESTRELLA—Head of Marketing @ Clay:
Thank you, Jordan. Great to be here.

Introduction — Why a second edition

host—JORDAN CHENEVIER-TRUCHET:
April, there's a big thing happening for you in 2026: you launched the second edition of Obviously Awesome, seven years after the first one. What motivated the rewrite, and what changed in your thinking?

guest—APRIL DUNFORD—Founder @ Ambient Strategy:
A bunch of stuff. The fun thing about doing a second edition is you get to do a do-over. When I launched the original, I assumed it wouldn't sell many copies — I had a very specific audience in mind, mainly B2B technology founders. Then it ended up selling way more than I expected. And when you sell a lot of copies, people send you emails and tell you what you did wrong. After a couple of years you see the pattern: places where I didn't explain myself well, or didn't spend enough time on a concept.

Three things really motivated the rewrite. First: handful of patterns I saw over the six or seven years after the book came out that I didn't fully understand when I wrote the first edition. Second: multi-product positioning. Up until the first edition, the vast majority of companies I worked with had a single product. After the book came out I worked with a lot of multi-product companies, and there are very specific issues there. Third: the process itself. I had broken positioning into five components but ten steps. Tech founders pushed me on that — eventually I realized some of it was pre-work and some post-work, so I rewrote it as five components, five steps, with explicit pre- and post-work around it.

When to work on positioning

host—JORDAN CHENEVIER-TRUCHET:
For people who haven't read the book — positioning sounds niche but it's everywhere in a business. When is the ideal time to work on it?

guest—APRIL DUNFORD—Founder @ Ambient Strategy:
When we build a product in the first place, we generally have a positioning thesis: assumptions about competitors, differentiation, value, market. We build to that and launch. The first wave of customers either validates or invalidates that thesis. In my experience, the thesis was never 100% correct. Sometimes close, sometimes wildly off. We'd be getting traction in a completely different part of the market with a completely different buyer than we anticipated.

Where I landed: that first wave is the point. In B2B, you usually get those first customers through your network. They're taking a chance on you. Your positioning doesn't have to be razor sharp for that first wave — they're coming through anyway. Once that wave is in, you know where you were right and where you were wrong. Then you can really tighten the positioning. So the perfect time? While you're building, before launch, have a thesis. But don't take that thesis, tighten it up, and put it on the homepage. Keep it loose so you're not closing the door on anybody. Let the market pull you. Once the first wave is validated, then you tighten.

Signals that positioning is broken

host—JORDAN CHENEVIER-TRUCHET:
You've been vocal that positioning should be revisited in fast-moving markets. Most CMOs treat it like one-and-done at Series A. What signals tell a CMO their positioning is broken?

guest—APRIL DUNFORD—Founder @ Ambient Strategy:
When I ran marketing teams in-house, we did this twice a year. Twice a year doesn't mean we changed the positioning twice a year — it means we got a group together and looked at it. We checked four things:

• Competitors. Are we starting to see a different set of companies on the shortlist in our actual deals? Just because somebody launches doesn't mean they're competing with you. If they're showing up on shortlists, we account for them.
• Differentiated capabilities. Maybe what was differentiated six months ago isn't anymore. Or we shipped a release that opens new value.
• Value. Big things change customer sentiment. COVID hit and growth pitches died overnight; everyone shifted to "do more with less" and "extend runway." Now AI is reshaping what customers want to talk about.
• Customer perception. Has it shifted?

If the answer is no across all four — don't change the positioning. Simple as that. But if something big happened, assemble the team. Don't just rip it up. Investigate whether you need to adjust.

Don't position against ghosts

host—JORDAN CHENEVIER-TRUCHET:
A lot of companies are right now messing with their positioning weekly, mostly because of AI announcements from competitors.

guest—APRIL DUNFORD—Founder @ Ambient Strategy:
If your positioning is working and nothing changed, don't mess with it. When I ran teams, marketing would start changing things mainly because they were bored. I'd come in, find the homepage rewritten, and they'd say "we've been saying this forever, it's boring, everyone knows it." That's a bad reason to change. You're the marketer, you stare at the homepage every day. Your customers don't. Often the team gets bored at the exact moment the positioning is starting to hit.

Early 2025 I worked with an ITSM company doing close to $100M. We did the exercise: "if you didn't exist, what would the customer do?" The head of product said: "Vibe coding is going to wipe us out. We need Replit, Lovable on that list." Those products had been out two months. So I asked the head of sales: "Do customers actually talk about vibe coding? Has anyone said 'maybe I'll vibe code this'?" Head of sales: "Not one." We did the positioning without vibe coding. A year later, I had a check-up call. Vibe coding has blown up everywhere. In their market: still zero losses, still nobody asking. Doesn't mean it'll never happen. Means I don't have to position against a ghost.

The competitor-panic trap

host—JORDAN CHENEVIER-TRUCHET:
I see two patterns with my clients. One: competitor changes their homepage and they panic. "If they did it, they're right."

guest—APRIL DUNFORD—Founder @ Ambient Strategy:
There's a lot of bad marketing out there. Just because they're running a campaign or changing their site doesn't mean it's working. Where you actually need to pay attention is sales: are they showing up in deals? An old boss of mine used to say "don't get caught smoking your competitors' marketing." Just because they said it doesn't mean it's anything. They might change it again in two weeks because it's not working.

Look at competitors, sure. But what you see externally is not a reflection of the go-to-market strategy. You don't know their segmentation, where they're getting traction, where they're struggling. You only see it in sales. If they're showing up on shortlists in your deals, take it seriously. Otherwise, marketers get afraid of companies spending a lot on marketing — but lots of companies raise spectacular amounts, blow it on marketing, then flame out two years later.

40-60% of B2B deals end in no decision

host—JORDAN CHENEVIER-TRUCHET:
You've said 40 to 60% of B2B deals end in no decision. Most marketing leaders blame sales execution. You blame positioning.

guest—APRIL DUNFORD—Founder @ Ambient Strategy:
Sort of. There's data here. Matt Dixon put out The Jolt Effect a few years back. They studied 2 million sales calls during COVID, then tracked what happened to the deal. Their number was 40 to 60% depending on industry. Roughly 46% of B2B purchase processes end in no decision.

They could also look at why. Sometimes it was "the alternatives look too hard, I'll stick with what I have." But mostly: indecision. Either the solutions all looked the same, or they looked scary. "What if I pick this and it doesn't work? What if I get in trouble with my boss? What if adoption fails?" In a lot of cases the company even says "yes, we picked you, let's go" — and then ghosts you.

If indecision is "they all look the same," that's a positioning failure. Get sharp on differentiated value. But also: the customer needs to feel confident they've done their due diligence, including on competitors. So make it clear "here's what they do, here's what we do, here's when you should pick them vs us." That feels counterintuitive but it builds confidence. Then risk — especially deployment risk in enterprise software. What does professional services look like, training, hands-on after the deal. Sometimes it's packaging and pricing: money-back if not deployed in X months. Sometimes you break the deal into phases: start small, prove yourselves, expand. You take the scary "all at once" off the table.

Often when customers say "price is too high" they really mean "this is too risky." So I don't drop the price, I drop how much they have to commit to right now. I worked at a company selling software to retail stores. Instead of selling 30,000 stores at once, we'd pilot 5, then 100, then region by region. Customer is much more comfortable because nothing goes wildly off the rails on a giant deal.

PLG vs Enterprise positioning

host—JORDAN CHENEVIER-TRUCHET:
The wall between B2B and B2C feels thinner now — PLG, self-serve, individual buyers. Are they converging?

guest—APRIL DUNFORD—Founder @ Ambient Strategy:
No. Lots of companies have PLG. Fine. But there's a big difference between an enterprise deal and an individual contributor deal. If I sell a $150 widget for a single developer and they slap down a credit card, that's not a complex sale. Low risk. If it doesn't work they buy another one next week. That's closer to B2C. One person, no committee. PLG often looks like that on the front end. Then if a few people in a team start paying, the company comes in over the top and sells an enterprise license. That's a totally different sales process and requires totally different positioning.

On the PLG motion, my positioning is "try it." I'm not running a competitive analysis — I'm getting in front of a person with pain and saying "have a look." On the enterprise side, I'm coming in saying "you should standardize on us." There are teams using other things who may not be happy about that. So I have to articulate the value of going from a few seats to every developer. Different positioning entirely.

Self-serve in enterprise mostly works for two situations: front-end discovery (figure out if there's demand), and cross-sell/upsell after the enterprise deal is closed. Nobody wants to talk to a rep when they're just buying 10 more seats of a database they already own. When I was at IBM, the bar was: under $250K of additional spend on an existing relationship, you weren't allowed to talk to a rep. Different beast.

Multi-product positioning — personas, segments, suites

host—JORDAN CHENEVIER-TRUCHET:
You added a section on multi-product positioning. Most scale-ups I work with have 2-4 products and end up with a Frankenstein: one positioning per product, per persona, per segment.

guest—APRIL DUNFORD—Founder @ Ambient Strategy:
Start with personas, because people overcomplicate this. In a typical B2B deal there are 5 to 11 personas involved. CRM example: head of sales ops is the deal champion. Then security, compliance, end users, IT, economic buyer. When I was a junior product marketer I was taught to do positioning for each persona. We did all that persona work (Marty doesn't like talking on the phone, etc.) and we never used any of it. We never used any of the personas except the champion.

Why? In positioning we are trying to get on a shortlist and get past the first meeting. The champion is the first meeting. The champion makes the shortlist. Our positioning needs to resonate for the champion. The champion might know they have to handle objections from security or end users — we help them handle those. But for everyone else, there's no "value for me." There's only objections. CSO doesn't care about value, they care about "is it SOC 2, are we going to jail." End users care about "do I like using it." That's objection handling, not value positioning. We arm the champion to handle the rest.

Then segments. We don't need separate positioning for retail vs utilities if the value prop is actually the same. We may put different emphasis on different value props in different markets, but the positioning is the same. I worked at a company selling to retail and utilities and the value was the same: business continuity. In utilities, "keep the lights on." In retail, "keep the cash register on." Same positioning. Different customer example in the pitch.

If you're early stage and the positioning really is fundamentally different across two segments, that's usually a sign you need to choose one market. You can't make both happy in the future. Choose.

Now multi-product. Position the way you sell. A division of a big company called me: five products. They wanted positioning across all five. I asked: how do you sell it? They always sold a data warehouse first and then layered analytics tools on top. So position the data warehouse, treat the rest as cross-sell. Another company had five products packaged like a suite — customers picked 3 or 5 or 2, but the mix changed every time. Position the suite: umbrella value. Sometimes it's a single landing product with cross-sell. Sometimes it's a suite. Sometimes it's a portfolio at the IBM scale, with layered positioning at the company, group, and product line level. The question is: what do you want the salesperson to pitch in the first meeting? Position to that.

Salesforce is a good example. For years their ads were "the world's #1 CRM." Land with CRM, cross-sell everything else after: Service Cloud, Marketing Cloud, Slack, Tableau, MuleSoft. Now they're shifting into a more portfolio-style positioning because customers might want to land on Slack or Tableau, not CRM. That shift requires a different kind of positioning conversation.

Category creation is mostly magical thinking

host—JORDAN CHENEVIER-TRUCHET:
Last big question. Almost every company that IPO'd recently competes in existing categories. They don't create new ones. But every VC-backed startup I talk to wants to create their own category.

guest—APRIL DUNFORD—Founder @ Ambient Strategy:
Category creation is a little bit of magical thinking. Wouldn't it be great if we could invent a name and then dominate because we invented it? If that worked, we'd all be doing it. What does a market category actually do? It tells the customer at a macro level what you're about. It gives them a starting point: who you should be compared to, roughly what you do. It doesn't replace your messaging or your pitch.

Positioning in an existing category is much easier because the customer orients instantly. Snowflake's early line was "a data warehouse for the cloud." We know what a data warehouse is. We know Oracle's data warehouses weren't born in the cloud, so we assume something special here for cloud workloads. Elegant. Easy. If I come in and say "I'm a flim-flommer," now I have to spend 20 minutes explaining what a flim-flommer is before I even start positioning.

There are very few examples of small companies that set out to create a category and then dominated it. Categories do emerge — especially in AI right now — but emergence isn't the same as a single company "inventing" a category. And historically the company doing the hard work of creating the category often doesn't survive to own it. We're not on MySpace, we're on Facebook. We're not using Ask Jeeves, we're using Google. We're not on BlackBerry, we're on iPhone.

Marketers also do wiggly semantics. A company told me "we created a brand new category." Me: "What is it?" Them: "CRM for lawyers." That's a CRM. That's a niche. A subcategory at best. You didn't invent CRMs. Where category creation actually works is when you're already very big and dominant in your existing category and you want to expand the borders. Then you redefine the market you already own to be a bigger market.

Investor positioning vs customer positioning

guest—APRIL DUNFORD—Founder @ Ambient Strategy:
Then investor vs customer positioning. If you have a SaaS product, your investors may be worried that 10 years from now people will vibe-code it. That's a fair worry. Your investor positioning may need to address it. But that doesn't mean your customer positioning has to. None of your customers want to vibe-code it. None of them are asking. You sell to them on what they care about today. The investor wants a 10-year story. The customer is buying today's dollar for today's outcome. Big difference.

The trap I see in workshops: the founder has been pitching investors for three months. They've been told over and over "what about vibe coding, what about category, what about the 10-year story." Then they're back at the office and that anxiety leaks into the customer pitch. It's natural. The job is to consciously hold the two apart. It is perfectly okay for the investor positioning to be very different from the customer positioning.

Where to find April

host—JORDAN CHENEVIER-TRUCHET:
Where do we send people who want to follow your work?

guest—APRIL DUNFORD—Founder @ Ambient Strategy:
aprildunford.com. Best place to start is the books: Obviously Awesome, brand new 2026 edition. Then if you want to go deeper there's a newsletter, a podcast, and a bunch of other stuff. All linked from the site.

host—JORDAN CHENEVIER-TRUCHET:
Perfect. Thank you, April.

guest—APRIL DUNFORD—Founder @ Ambient Strategy:
Thanks so much for having me.

Introduction

host—JORDAN CHENEVIER-TRUCHET:
Hey Kieran, welcome to the show.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
Hey, thanks for having me on. Excited to be on.

host—JORDAN CHENEVIER-TRUCHET:
Kieran, I think we have a lot to talk about today, but mostly around AI — because you've been quite loud on AI, the way you use it, your thoughts on it. I really want to dedicate this episode to that. What's interesting in your case is that you have the two levels: the strategic one (AI marketing leader), and the operator side. Especially coming from your background as a software engineer. You were a software engineer, correct?

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
Yeah, I was a failed software engineer. I did computer science at university, did some years coding — Linux, Unix, Perl, Python. Going back to the command line through Claude Code has actually been great for me. I love working through the command line.

Killing the middle tier of marketing orgs

host—JORDAN CHENEVIER-TRUCHET:
You've said publicly that AI is killing the middle tier in marketing organizations — people who aren't exceptional craftspeople or highly technical. The thing is, most CMOs and VPs of Marketing have built their org around exactly that middle: headcount, job levels, team design. So how should they consider changing the marketing organization?

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
I was speaking to a founder recently who uses a lot of agents in his business. His take is probably right for where we are today: if you have someone who's mediocre at their job, agents can take that work and do it just as well. They cannot do the work of a true craftsperson, a real domain expert. Maybe over time they will, but that's where we are today.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
How does it apply to marketing? What happens in all organizations is that we flatten the org structure — people become much more autonomous and can do much more work because of AI. Marketing is unique among go-to-market teams. In sales, sellers all have pretty much the same role. SDRs, BDRs, AEs — same with customer success, support, even engineering. Marketing has a whole slew of niche teams: product marketers, brand marketers, demand marketers — all with their own domain expertise but not really cross-functional.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
What I think happens because of AI: you'll have more generalists who can do all of marketing, with agents and AI as part of their team. The org gets flatter. Probably less middle managers. People managing many more individual contributors. Those individual contributors will be generalists with agents and assistants helping them across all facets of marketing. And that's already happening — spend any time with AI-native companies, that's how they're set up. Not like traditional marketing teams in tech.

The most dangerous employee right now

host—JORDAN CHENEVIER-TRUCHET:
For most marketing leaders, they ask their team to adopt AI to be more productive — but people internally start to wonder, "am I just training AI to replace me?" How do you manage that tension?

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
There's always going to be a world where you need to have great taste. The most dangerous employee within any organization right now is someone with a small amount of domain expertise and a high amount of AI agency. They're not very good at their job, but they're really passionate and shipping a lot of AI work. They're the most dangerous people in any organization, because they're creating a real amount of slop.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
We're really in the messy middle right now. AI has given people superpowers — they can ship code, ship memos, ship an infinite amount of stuff. But without real domain expertise and understanding why you're using AI, what outcome you're trying to drive, it just creates more and more noise. Everything looks like an AI problem: "I'll use AI for this, ship more of that." But are you really driving more outcomes? Are you really making the business more successful?

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
The best people in a company are the people who have real taste, real domain expertise, and high AI agency. They're the ones creating most of the momentum today. As for the question "am I training AI to replace me" — AI is more on personal accountability than the company telling you to do something. It's your choice to make. AI is going to happen whether any of us like it or not. We have very little control over it. So all you can do is try to make the best choices to be the best version of yourself in a world where AI is prevalent.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
What I'd be thinking about: in a world where AI and agents augment what I do, what is the real skill, the real craft, the real domain expertise I can bring to make myself valuable? And one more thing: model capabilities are moving really rapidly, but integration into how companies work is moving really slowly. There's a lot of opportunity to be the one who figures out how to integrate AI into your team, into your business. That's a skill set most companies need.

Why AI productivity is harder to measure in marketing

host—JORDAN CHENEVIER-TRUCHET:
You've said AI projects live or die on data quality. You shared that at HubSpot you had an AI personalization workflow for booking meetings, and it took 3-6 months to get the prompt right. Most B2B companies have fragmented data, messy CRM. What's the minimum viable data foundation?

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
Every time we've improved the quality of our data feeding the AI to personalize emails, chat, anything — the increases in conversion rate have been pretty startling. Every time we integrate a new dataset and unlock more personalization, we see a 3 to 500% increase in conversion rates on email, which has been around forever. The minimal viable version is whatever you need to elevate your AI experience above the average in your industry. Data sources depend on the vertical and your customer.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
You want a data repository with clean, enriched data the AI can easily use. The example you mentioned was AI prospecting — we've done that all the way through the funnel. AI has taken over a lot of our prospecting and we've seen huge results. Each time we've given the AI a better dataset — internal CRM data, external from various sources — we continue to see real performance increases.

host—JORDAN CHENEVIER-TRUCHET:
You mentioned earlier that measuring AI in software engineering is easy — you have lines of code, you can correlate that to productivity. In marketing, it's harder. What do you follow to see you're on the right track with AI?

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
It's a good question. I don't think we've come up with a de facto way to measure all the different teams. We do show-and-tells, we have a singular Slack channel where people post what they're doing with AI integrated across go-to-market. There are three different ways AI usage shows up: my personal usage, my team usage, and how I integrate it into the customer experience. We measure a lot on the customer experience — AI prospecting, sales agents, support agents, success agents — based on business performance.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
The harder one is visual usage. If a product marketer is using AI, what should we expect in terms of output? Five positioning statements per month vs twenty? It doesn't make sense. This is why I think if you're going to lead teams through this next iteration — and you're best in marketing — you have to be an AI-native leader. There's so much nuance in understanding why and how your team is using AI, whether it's driving the right outcomes. There's no clean metric. You yourself have to really understand AI, dig into the details, and provide good feedback. Marketing more than probably any role should have leaders who are quite AI-native, probably in the weeds doing a lot of stuff themselves.

How to hire AI-native marketers

host—JORDAN CHENEVIER-TRUCHET:
How do you make sure when you recruit people that they're already using AI the right way?

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
You can give people tasks, and you make the task pretty impossible unless they actually understand how to use AI. In the past, you'd hire a product marketer with a positioning statement exercise. Now you can say "rebuild the website" as part of the task. That's a really easy thing to do with AI. And if they rebuild the website, you see how they bring positioning to life across pages.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
You can give tasks that are a forcing function — if you don't know how to use AI for this task, you'll think it's an incredibly impossible thing to do. Within the interview process, you can also talk to someone and understand whether they use AI. It's not something you can fabricate if the person on the other end actually knows how to use AI. So: the test you give, the interview process, and the portfolio of work — show me real examples of problems you've solved, times where you had high agency and used AI to solve problems traditionally handed off to others.

Creator-led marketing & AI mascots

host—JORDAN CHENEVIER-TRUCHET:
You've said creator-led channels are becoming the dominant B2B brand asset. But most B2B companies don't have a Kieran Flanagan at the top — their executives are cautious about going public. What should they do?

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
Setup: in 2021 we bought The Hustle. We thought B2B would look like B2C over time — usually B2B is 3-5 years behind B2C. Back then, B2C attention was shifting into creators and individuals. Branded channels like paid CAC were going up. Blogs were getting disrupted by AI. The channels growing were personality-led — YouTube, newsletter, TikTok, Instagram — where the person is at the forefront. We weren't really thinking about AI back in 2021. We bought The Hustle and built a media network — podcast, YouTube, all assets led by people versus brand, but part of the HubSpot network.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
That's been accelerated with AI. People want to connect with people because they trust people, and a lot of branded content gets lost in the AI ether. There's just so much content out there. We're seeing that happen in B2B today. Your question — can every company do this if they don't have a good creator within the org? It depends on the vertical and the customer.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
Here's what I think will happen: most brands will have an "AI mascot." That AI mascot will be the face of their brand. An avatar — a synthetic — and people may hate this, but it will be a synthetic avatar that looks very lifelike. It will bring your brand to life across these personality channels. Your content team is the brains behind that mascot. You might think that's a wild idea — it's already happening in B2C. A lot of the biggest Instagram accounts and the fastest-growing ones are AI avatars. You can't tell the difference, but they obviously disclose it. Same on X. You always want to disclose it.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
Will people consume content from AI? Already proven — they are, whether they know it or not. There was a study released — I think NYT or another large media org — a blind study where they gave people AI content and human content. The AI content won. Humans objectively, when they don't know it's AI, prefer the AI content because it's better than the average human content. Will there be a moment of backlash? Probably. But it's going to happen. Pull ahead three years and I suspect that some of the biggest creators in B2C and B2B will be avatars with teams of humans behind them.

Kieran's daily AI workflow

host—JORDAN CHENEVIER-TRUCHET:
Before you started recording, you said showing your day-to-day AI use cases is like asking you to show your day-to-day computer use — hard to even know what to highlight. Can you elaborate?

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
Early on when I talked about AI on podcasts, people asked: "show me what you do in Claude." Back then, lots of ChatGPT usage. I've migrated most of my usage into Claude. I'd say things like "here's a memo, red team it, give me all the reasons it won't work" — reverse engineering, like Amazon does. Then you fix problems before doing the thing. A lot of strategy work. I had an agent doing first-principles thinking, then frameworks. But over time, the models got so good I don't think about how I'm using it anymore. It's just part of how I work.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
One tip I can give: I have a Claude project for everything I'm working on. Every document, artifact, meeting note goes into that project. I think of them as mini brains. For a project, I have a mini AI brain in a Claude project. I go to Glean (we use it internally), pull all my notes, meeting notes, Slack messages — and feed them to Claude. We've connected to Glean's API so I can say "before you answer, retrieve the latest updates in these docs." Everything I can put in, I put into Claude in that project as a little brain. It helps me work through things in such an expedited way.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
Other tips on tools: I built all my decks in Lovable. Recently I did a pitch off internally with a Claude Code-themed presentation styled like a terminal. Lovable lets you create any style. Claude is also good for presentations. NotebookLM is great. Anything using NanoBanana works well. I've passed the part where I have defined workflows. Instead I'm just working on a task and Claude is a partner working on it with me.

host—JORDAN CHENEVIER-TRUCHET:
You generated the deck in Lovable in terminal style — but can you really get these tools to do presentations linked to your brand book?

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
Yeah. A UX designer did an awesome internal presentation recently for a leadership team. I asked Lovable to copy it for me. One shot. Completely replicated the design. If you give it your brand styles, colors, palette, it can replicate your presentation really well. The reason you might not use Lovable for external conferences is you'd need a PDF. But anywhere you can do something using a website for your presentation, the creativity in Lovable is great.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
One thing I'd push on: should we even be doing things the way we've always done them? All our decks are uniformed, same brand styles and templates. Maybe that's not the way things should be done. AI lets you be creative in ways you couldn't before. What I see a lot of people doing is "how do I get AI to replicate the things we did before?" That's the worst use I see — "enter your domain and we'll replicate your writing style." But your writing style is probably average. Why would you want to replicate it? AI can do a hundred times better than what you do. Why force it to do something that's bad and generic? I don't think people use AI in the right way. They try to do more of what they already do, not elevate what they do.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
We're in the era of show, don't tell. Anytime you can prototype something and show it, it's way better than a presentation. Way better than a memo. Not always doable, but trying to think about how AI can let us do something different — not just replicate what we've done before with AI faster.

Live demo: Kieran's Claude Code content system

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
Let me show one of the systems I built in Claude Code. The content team works automatically — I give it a content strategy and it creates a certain amount of content each week. Lessons that apply to any model, any chat: this works just as well in Claude desktop.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
The team is split: an onboarding process creates an audience profile, then a writing style. The writing style can be based on anyone — give the team a name and it creates a style. If you're not a great writer, don't tell it to recreate your style. Tell it to elevate your style.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
People say "this is copying others' writing styles." There's a great copyright framework called copywork from the early 1900s — that's how people got really good at copywriting. Before computers, they'd literally take great writers' work and write it out by hand, letter by letter, to get used to the beats of how someone creates great content. Then you enhance with your own style. That's the writing-style step in my system: get a style card as a foundation, then build your own voice on top.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
Then research skills. They create a queue, stack-ranked. An agent provides feedback to the queue and constantly improves it. I think of content in modular blocks: a hook, an intro, an enrichment skill, individual skills per platform. The big thing if you're building real systems: it has to have feedback loops. A performance agent scrapes the web, looks at the content created through the system, integrates analytics, and another agent updates the entire system based on learnings.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
An orchestration skill ties everything together. Rather than remembering all the skills, I just say "let's create content" and the orchestration skill knows all the others. It feels like a little app.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
What's it doing? It's connected to X, Perplexity, OpenAI, and Firecrawl (a tool that crawls websites). Audience profiles are based on what works for me on each platform — the system looks at all my Substack posts plus analytics and says "based on that, this is a winning profile." That matters because the research skills then go find ideas that look like the winning formula for each platform.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
Lessons for anyone building this kind of thing: (1) Context really matters. The profiles get pre-loaded — the system has context on who it's creating content for. Whether you're doing product marketing, content, anything — context drives quality. (2) Prompts matter — and they're based on your domain expertise. Your ability to tell the system how to complete a task really well is what makes it skillful. (3) Feedback loops. Show the system examples of what you wanted vs what you got. It gets much better when you're clearer about the outcome you want.

host—JORDAN CHENEVIER-TRUCHET:
How long did it take to build this system?

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
It's based on an app I launched last year. The prompts behind the system took about 9 months of iteration. Building it inside Claude Code wasn't long — coming from an engineering background, system building feels logical. Claude Code is so quick at building systems. There's a slash-insights command — run it and Claude shows recommendations for improving the project. You can just say "implement these" and it does.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
Over time you can ask Claude Code: "based on the system we've built and your improved capabilities, what should I change, what don't I need anymore?" It really helps you simplify. The value I add is domain expertise — how to bring this to life in a real content system that adds real value. Claude can build the skills and scripts. I didn't build any scripts. It built all of them.

host—JORDAN CHENEVIER-TRUCHET:
How clear are you on the destination before involving Claude? Or do you involve Claude in defining the destination?

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
I start with an idea of the system, then work with Claude. I always say to Claude "are we aligned?" — that's a great thing to do before it starts working. It does the ideation. In this example, I had a ton of source code from the app and gave it the entire source code: "here's what I have. How can we replicate this into a system?" It extrapolated all the prompts to start building skills. Then it's iteration. I added the content queue because the more research it did, the harder it got to figure out what ideas it had created. I added feedback loops because it wasn't always right — needed guidance.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
I'm rebuilding the YouTube part because YouTube is so different from text. It's involved in the entire process from start to finish — sometimes Claude is the dominant partner taking the lead, sometimes I'm dominant telling it exactly what to do. Not too different from how you work with an engineer.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
How do I use the output? I never copy and paste content. I create ideas enriched with modular blocks — a story, a case study, an authority quote, a counter-argument, a visual hook — and then I rewrite from that. The system is very much built to how I work. Now if I were giving it away, I'd add an onboarding agent that adapts the system to the user. AI is very personal, very nuanced to how you want to work.

Where to find Kieran

host—JORDAN CHENEVIER-TRUCHET:
For people who want to follow you and learn from you, where do we send them?

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
I did a LinkedIn post about my AI content system. I literally said as a joke "you can only get it on my Substack." I now have 40 DMs on LinkedIn saying "AI content system." So stop DMing me — you can get it on my Substack: kieranflanagan.io. And our YouTube channel is Marketing Against the Grain — with my friend Kipp Bodnar. You'll see a lot of this stuff there visually, watching us build it.

host—JORDAN CHENEVIER-TRUCHET:
All right, thanks a lot Kieran. Talk to you soon.

guest—KIERAN FLANAGAN—SVP Marketing, AI & GTM:
Thank you.

Introduction

host—JORDAN CHENEVIER-TRUCHET:
Hey Alice, welcome to the show!

guest—ALICE DE COURCY—ex-CMO Cognism:
Hi, thanks for having me.

host—JORDAN CHENEVIER-TRUCHET:
Alice, you're the ex-CMO at Cognism. Can you walk us through your journey at Cognism?

Joining Cognism: From €3M to €100M

guest—ALICE DE COURCY—ex-CMO Cognism:
I was at Cognism for nearly seven years. I joined when we were just under three million in revenue. I remember our €3M revenue party, probably about a month into having joined. When I left, fast approaching the hundred million mark.

Across that seven years, it was an amazing journey. Leading the marketing function from a team of three when I joined to about 28 of us in the marketing organization, operating globally across core European markets and the US. We went from a very SMB-focused offering to a consistent revenue split across small business, mid-market, and enterprise.

Company size: we went from maybe 25 people total to at one point 450. We acquired a couple of brands during that time. The reason I was brought on was to create a predictable, scalable, revenue-generating marketing engine. That was the job for the last seven years.

What Changes From Build to Scale

host—JORDAN CHENEVIER-TRUCHET:
What was the biggest difference between three million in revenue and almost a hundred million?

guest—ALICE DE COURCY—ex-CMO Cognism:
When you're at zero to one or three to four, you have to find out how good you are at problem-solving and building things with little to no resources. You learn things that traditionally wouldn't fall into your role, because there isn't a role for that.

You build operational infrastructure from the ground up. You'd never do that at a larger organization, but you absolutely have to do when you're in those early days. Can you troubleshoot problems no one else is owning? Are you willing to get stuck in?

You get good at unscalable things in early stages because that's what gets success. You operate fairly blind on data — there's only the foundations you build. I'd built for the fact that we would scale, thinking through "will this break when we get further?" Worth putting time into building a foundation. You have to be comfortable acting on directional insights, not full datasets.

And then it was hard for me to move more hands-off as we got bigger. The expectation is that you're no longer the bottleneck to scale. You need to give that knowledge and repeatability to your team.

Becoming a First-Time CMO: Re-Interview Yourself Every Year

host—JORDAN CHENEVIER-TRUCHET:
You were hired as Head of Marketing, then became CMO. How did you build the trust to become CMO?

guest—ALICE DE COURCY—ex-CMO Cognism:
First: don't have unrealistic expectations. This doesn't happen overnight, especially when you're a first-time CMO.

The way I thought about it: every year I was reinterviewing for my job. Whether it was Head of Marketing or CMO, every year I needed to re-interview for the role. Nobody told me. Nobody put me in that process. It was just my mentality, because I had never done that next phase. So having that as my foundation was helpful.

Building Trust With the CEO

guest—ALICE DE COURCY—ex-CMO Cognism:
Second: I worked out what mattered to the CEO. For every CEO it's different. I was fortunate — working for an inspirational founder-CEO who was passionate and clear about what he cared about. Part of the skillset of a CMO is working out what matters more than others.

I used to say: I know something really matters to James (the CEO) when he's mentioned it multiple times in the same time period of a day. Not within a week — within the same time period of a day. That meant: highly critical, prioritize getting an answer.

I worked out how he wanted to be communicated with, where he'd get frustrated. He was a founder-CEO operating at hugely high intensity. He wanted people to operate at that level with him. Learning quickly: when he asks me a question, give him an answer in the moment. That was where I needed to meet him.

Always have the numbers. Same reason as the communication piece — if he needs an answer, it's because there's a piece of work, a board member, an investor. Being on top of your numbers all the time, not having to come back or find it out from someone else, was important. Helped build a trusting relationship quickly.

Marketing the Marketing Internally

guest—ALICE DE COURCY—ex-CMO Cognism:
I wasn't very good at this in the early years, but I got better later: be super proactive distilling down what marketing was doing and how it led to the overall growth.

What I needed to be less afraid of was repetition. Repetition is what builds memory. The more I proactively communicated marketing's role in our growth and wins that mattered strategically, the better. He wasn't going to come and ask. It had to come from me.

host—JORDAN CHENEVIER-TRUCHET:
Does this apply to all CMOs?

guest—ALICE DE COURCY—ex-CMO Cognism:
It applies to anyone in a job who has a boss. Understand what matters to them. Communicate effectively. Proactively understand how what you do helps the overall goal. Be on top of your numbers. Repeat and internally market. Regardless of CMO title, those concepts are true.

Building the Team: Don't Hire to Hire

host—JORDAN CHENEVIER-TRUCHET:
How did you build your team?

guest—ALICE DE COURCY—ex-CMO Cognism:
I was big on: don't just hire to hire. I needed something to break or nearly be breaking before I'd hire into that role. Second, we were resource-constrained. I wasn't hiring people for tenure. I was hiring for attitude and ability to understand the new playbook and philosophy I wanted to build.

I had a core belief and understanding of what kind of marketing organization I wanted to build. People are a good or bad fit for that. Quickly. We trained up people at lower experience levels, more generalist. As things scaled, they became more specialist where the data showed we should invest.

The Structure at €100M Scale

host—JORDAN CHENEVIER-TRUCHET:
When you left, what did the marketing organization look like?

guest—ALICE DE COURCY—ex-CMO Cognism:
Paid acquisition function for paid channels. Demand generation — my business unit owners. They owned targets per region or segment, fully responsible for revenue and pipeline. Other parts of marketing acted as mini specialist agencies to them.

Content and SEO were split: SEO for capturing demand (organic search and LLM search). Content for building the demand creation media engine.

Customer marketing and product marketing under the same VP layer. Then the enablement team — video, website, design — a resource across the whole org.

Critical at scale: an incredible project manager who kept every cross-departmental project on track and built processes around it.

Where AI Would Have Changed Things

host—JORDAN CHENEVIER-TRUCHET:
What would you have changed if you had AI tools when you were at Cognism?

guest—ALICE DE COURCY—ex-CMO Cognism:
Two categories of mistakes I made. First: when to in-house and when to outsource. We went through a growth-at-all-costs era where we just threw people at problems. The problem: you can only keep quality of hires to a certain level when your team is a certain size. At too large a scale, you make sacrifices. The overall ability of your team diminishes with scale. Hiring more people doesn't mean more scale — it just adds complexity, silos, training requirements.

Two concrete examples: paid and SEO. I had a strong belief I wanted all the experience and domain expertise in-house. Still important. But the cost of managing that number of people — ensuring career pathways, work variety, engagement — it can't happen at scale efficiently. There's a ton of execution work that's really boring. AI is taking a lot of that away.

Specialist in-house — your strategic knowledge and domain expertise — plus an agency or AI for execution. That's where I ended up. You execute at scale efficiently. You outsource the HR and career development issue. You don't lose internal domain expertise. People get rewarded more appropriately for strategic work. Had AI been where it is today back then, that would have happened sooner.

The Ten-Years-At-Salesforce Trap

guest—ALICE DE COURCY—ex-CMO Cognism:
Second mistake: senior hires. My philosophy was to grow people within the organization. That got challenged as the C-suite tenure changed. I'd built a marketing function with clear attributes and beliefs, not based on years on the CV.

I was open to experimenting. It comes at high cost to your P&L, hiring with much higher experience levels. The promised land: it'll be better, you'll manage less, you'll be less in the weeds, you'll learn from someone better at that domain than you.

A good example: product marketing. We tried a couple of these hires — with no success. Never lived up to expectations or cost.

Just because someone has tenure and numbers of years on their CV does not mean they'll best execute that role in your organization. Done at scale, it's detrimental to culture. People see they're hired over with the promise of learning from someone more experienced, and if that value isn't delivered, it's just an additional layer between you and the work. It's a career blocker, more expensive, less efficient.

host—JORDAN CHENEVIER-TRUCHET:
When does it make sense to hire someone with 10+ years from a big company?

guest—ALICE DE COURCY—ex-CMO Cognism:
I wouldn't. The only exception: someone like that who went to a build-phase company, stayed there two to three years, succeeded, then went back into building. I don't think hiring someone from that enterprise type and putting them into a build phase if they haven't been in build phase for 10 years is ever going to be successful.

You really differentiate the manager versus the builder. We were never at a scale where anyone could just manage. I don't know if any company can afford to just have people manage now. Everyone has to be operational and delivering unique value.

Why Most First-Time CMOs Get It Wrong

host—JORDAN CHENEVIER-TRUCHET:
What do you see first-time CMOs consistently get wrong?

guest—ALICE DE COURCY—ex-CMO Cognism:
Too hands-off for too long. CEOs say to me on a weekly basis: "I just want to get some stuff done. I want to get into execution mode."

That's a critical piece marketers are shying away from — the execution work. There's a pandemic out there: people are scared to hire marketing leaders because all they think they're going to get is a positioning statement, a bit of messaging, but nothing that moves the engine.

Tied to this: it's critical to set expectations upfront with the CEO about what they're trying to achieve with a fractional hire on board. Make sure you're aligned on the output. It's not about buying hours — you should be buying an output. I should be delivering a product that helps your business move forward.

The Math That Killed Gated Content

host—JORDAN CHENEVIER-TRUCHET:
How do you sell demand creation to a C-suite?

guest—ALICE DE COURCY—ex-CMO Cognism:
The Cognism story. I was a full-on advocate of ebooks, gated content. The best company at generating them at the best price. I took Chris Walker on a podcast and challenged all his beliefs. I was committed.

Then we got to another phase of growth. I was doing the math. I split the funnel: declared intent demo requests on one side, content leads on the other. Conversion rates from demo request to closed-won versus content leads to closed-won.

It was super stark. I needed 25 demo requests to close 1 deal versus 100 content leads to close 1 deal. On top, ACV decay on content leads — half the ACV size of demo requests.

When you do that math, look at next year's targets, think about budget and resources — not just on marketing, on sales to follow up — the math doesn't math. It becomes unscalable.

I knew the first question was: "can't we just do more demo requests?" Yes, but how do I do it scalably and repeatably? That came from building a truly engaging, authoritative content engine that acted like a media machine — always on, all the time, to everyone in our ICP, delivering differentiated value. I couldn't do that AND gate content. Those two things didn't exist in the same world.

Selling Demand Creation to a CEO

guest—ALICE DE COURCY—ex-CMO Cognism:
I started showcasing the data and showing the inefficiency of the old way. At the same time, sales didn't love calling content leads. To have them convert one in 100 times — they didn't enjoy that. I surfaced that feedback.

The thing that really started to work: I asked the C-suite, "how do you buy?" When you see an ebook, do you give your real personal details or use an alias? You know it's coming — a PDF in your inbox. Does it actually get read? You know you'll get cold-called from an SDR. Does that give you a great feeling? Are you more ready to buy? The answer was always no, no, no.

It's hard for people to go against the logic that this isn't the reality of how we should be consuming content. The math, the reality, plus surfacing that sales are aligned — those three things generated buy-in.

host—JORDAN CHENEVIER-TRUCHET:
You have investors. You need to show numbers. How do you make the math math?

guest—ALICE DE COURCY—ex-CMO Cognism:
This was always my argument back to Chris Walker. We're not going cold turkey. I'm going to turn down and turn off the most inefficient content campaigns. Repurpose that money into ungated content. And go to my CEO with a hypothesis showing how I'll scale demo requests in line with the spend that fills the gap on content leads.

The thing in my advantage: the crazily terrible conversion rates on content leads. I didn't need many more demo requests to fill that gap in deals closed. By turning down inefficient campaigns and turning up the new way, within two months I had correlation graphs showing spend and demo request increase. That kept the buy-in.

Then it gets more sophisticated. As you scale, putting hundreds of thousands monthly into ungated content, you get more sophisticated. You invest in software — Fibula on the LinkedIn ads side. We ended up with Dream Data, which helps across the board on the journey before they're known to you. And finally we were building proper marketing mix modeling — that's the €100M+ mark.

Why Companies Fail to Pivot to Demand Creation

host—JORDAN CHENEVIER-TRUCHET:
Why do some companies fail at this pivot?

guest—ALICE DE COURCY—ex-CMO Cognism:
It can be made to sound very simple: "you just ungate your content and demo requests go up." Some people aren't getting it ungated and they expect demo requests to go up. That's not what I'm saying. There's a lot of nuance.

The Four Content Buckets

guest—ALICE DE COURCY—ex-CMO Cognism:
You need a strategy driving the create demand engine. We sat down and thought: we need topics we want to be known for. Memorability is built through repeatability. Random acts of content, even ungated, won't have impact. The four content buckets we covered:

• Thought leadership — content with nothing to do with our offering, but the new story of the day for our industry. We should consistently break it in a helpful, actionable way.
• Dotted line back to the offering — we still need people to know what we do. We were a B2B contact data provider, so topics around cold calling, objection handling sat here.
• Product content — not traditional product marketing decks. How does our product solve a problem? Show it through the voice of a customer or third party.
• Social proof — ways to validate what we're saying with facts and opinions of others, not just us.

And distribution. If you ungate a PDF and put it on the website, no one knows it's there. Understand where your audience hangs out, how they want to engage, what formats work for those channels.

You cannot cheat quality. That was always my bet. We're never going to be as big as competitors on spend. But we can create truly authentic, free quality content that an LLM can't write and competitors can't do at their scale, because we go to the best in the industry, get their insights, drive content from that. People don't do that. That's the other part.

Acquisitions and Brand Integration

host—JORDAN CHENEVIER-TRUCHET:
Cognism acquired companies. How did you manage brand and operational integration?

guest—ALICE DE COURCY—ex-CMO Cognism:
Pragmatic. Strategic reason for the acquisition drives everything. Take Caspr as an example — they had an incredible brand presence in France and we were going into France. We weren't going to kill the brand. We were going to maintain and utilize it.

It was a PLG motion, we were sales-led. Helped us decide we weren't fully integrating — we didn't want to cannibalize our SMB audience on a sales-based motion. Different brand for that reason.

On the back end, fully integrated. In Paris, Caspr and Cognism in the same office, same systems, integrated teams. The Caspr marketers became the French leads — experts in the territory. Front-end strategy by acquisition rationale, back-end fully integrated where it made sense.

Two Books and Closing

host—JORDAN CHENEVIER-TRUCHET:
You wrote two books. Give me the sales pitch.

guest—ALICE DE COURCY—ex-CMO Cognism:
There isn't much of a pitch. The honest ramblings of someone trying to master the CMO role for the first time. A story of two halves — the early growth phase is so different to the later stage. I had to interact differently with the C-suite, scale my strategic lens, get less in the weeds.

The journey from a content lead generation business into a fully bespoke create-demand motion driven by a systematic content engine. It came about because we were living what we preached — we should be a media company. Publishing books is what a media company would do. I'd been open-sourcing my learnings on LinkedIn the whole time.

It was a great opportunity to test whether this format could be repurposed and packaged for the audience. It worked, but it was for Cognism. Not done for me or my personal brand.

host—JORDAN CHENEVIER-TRUCHET:
The ICP is the future first-time CMO of a company?

guest—ALICE DE COURCY—ex-CMO Cognism:
Anyone in a leadership role for the first time, or someone who wants to be a CMO or is in a role for the first time and is building. There's lots that helps regardless of exact title.

host—JORDAN CHENEVIER-TRUCHET:
Where can people reach out?

guest—ALICE DE COURCY—ex-CMO Cognism:
LinkedIn. I'm very active. Put the bell on the LinkedIn posts.

host—JORDAN CHENEVIER-TRUCHET:
Thank you Alice.

guest—ALICE DE COURCY—ex-CMO Cognism:
Thank you.

Introduction

host—JORDAN CHENEVIER-TRUCHET:
Hey Dave, welcome to the show!

guest—DAVE GERHARDT—Founder Exit Five:
Thanks for having me, Jordan.

host—JORDAN CHENEVIER-TRUCHET:
Now everybody's talking about AI. CMOs and VPs of Marketing are scared. There's huge FOMO. The CEO and the board are pushing "how do we double revenue and divide the team by three with AI and agents?" On your side, you talk a lot about events, social, organic, storytelling. What's wrong with you?

The AI Hype vs Reality Gap

guest—DAVE GERHARDT—Founder Exit Five:
I'm not an engineer. This is the perfect time if you came up to marketing from being an engineer. My friend Tom Wentworth, CMO at Incident IO, posts every day about things he's building with Claude Code. He has engineering background — perfect for this era.

My career in marketing is more on the brand side, the product marketing side. So that's my bias. But I also think there's a lot of nonsense out there right now. I run a community of B2B marketing professionals — CMOs and marketing leaders. There's a huge gap between what you see on YouTube and what's actually happening inside companies.

If you go to YouTube and type "Claude marketing," you'll see 20 videos: "I replaced my marketing team with Claude Code in 20 minutes." Then the next one says 16 minutes. There's a huge cash grab and attention grab because of how viral all these topics are right now.

Someone who makes money writing AI workflows content is very different from what's actually happening inside existing companies. I'm not an idiot — there's also real stuff in there. If I was starting a company today, the makeup of the marketing team would certainly be different. But existing companies aren't going to fire 100 people in marketing overnight. Maybe in the next couple years.

I'm at the intersection of "I believe in AI, I believe in where things are going" but I also talk to lots of people every week who'd be surprised how many people aren't even using Claude regularly. Just using Claude is a revelation for them, never mind the use cases shared online.

And ultimately, you're still selling to people. Until my AI agent does research for me and buys on my behalf, exclusively talking to your AI agent, humans are still in the buying process. There's a lot in social, community, and events that pays towards humans.

Do You Use AI?

host—JORDAN CHENEVIER-TRUCHET:
Do you use AI?

guest—DAVE GERHARDT—Founder Exit Five:
Absolutely, of course. I am AI. Claude was actually trained on me.

host—JORDAN CHENEVIER-TRUCHET:
From the way you use AI, what do you think will be in danger on the marketing side, and what will never change?

What Stays Timeless: Human Connection

guest—DAVE GERHARDT—Founder Exit Five:
A lot of stuff that can be replaced makes sense. Advertising is a perfect example. It's always been about how many landing page variations, copy, creative, headlines, images you can produce. Even when I worked with a strong growth team and had a great idea, they'd say "cool, but we need 50 variations." Ad creation, landing page creation, automating ads — the future is absolutely going to be: I should just be able to go to LinkedIn, type in plain text "show these ads to these people," and it just happens automatically.

But on the other end, we just came back from doing this event with 100 people in Arizona for B2B CMOs. Highly curated, lots of offline time, real conversations — about work, family, life outside work. There was something there I can't perfectly articulate, but there's something about that human in-person connection.

We're not all going extinct in the next couple years. There's a fundamental need to still want to connect. There's a trend with my kids. Content creators are filming videos with old-school video cameras because they want that look. iPods are back. Landline phones are back. People are dying for a phone with no screen, doing retreats, "raw-dogging" a flight — just sitting on the flight without using your phone or laptop. We're trying to find ways to reconnect, be human, be present, be in the moment.

I want to be in both camps — believe in AI, use it to run my business, but not lose sight of humans since the beginning of time being connected over the campfire, telling stories. Unless you believe everything is going to be automated and your AI agent will buy from my AI agent at my agency, I'm in the middle camp.

Marketing's New Superpower With AI

guest—DAVE GERHARDT—Founder Exit Five:
What's exciting for me as a marketer: I became a CMO because I was good at marketing. You get promoted in your career, eventually rise to the top, and "congratulations, you're good at marketing — you now manage the whole marketing team." Turns out when you do that, you don't actually do a lot of marketing. You do people management, budgets. Most of my day was with HR, the CFO, the CEO, fundraising. I was like, "man, I just like writing copy on a landing page and making an ad."

What's so cool now: I can have an idea — I'm not a designer, my design sucks; I'm not a developer, I only know basic HTML — but I could go this afternoon and build my own website. The speed at which we can go from having an idea to making something is super exciting. I felt this way about Canva before AI: "I can make decent thumbnails for my videos by myself."

Dave the marketer can now do all the things in marketing from my video game station — my computer.

From CMO at Drift/Privy to Building Exit Five

host—JORDAN CHENEVIER-TRUCHET:
I felt that when you left Drift and Privy, you were at the top of your game. You had to start again from scratch — you built Exit Five and the community. How did you handle that?

guest—DAVE GERHARDT—Founder Exit Five:
I never felt like I went from scratch. The thing that led me to build Exit Five was that when I was doing marketing at Drift and Privy, I started writing about marketing on LinkedIn. I had a podcast and a newsletter. It was fun.

Tale as old as time: if you write about the things you're doing online, you attract the audience of like-minded people. That led to me unintentionally, accidentally building a large following on LinkedIn of B2B marketers. It wasn't because I wanted to be famous or have a personal brand. If you actually know me, I live in the woods in Vermont. I don't drive fancy cars. I don't really care what clothes I wear. I'm never going to be the personal brand guy in front of a Lamborghini. I like to hang out with my family, work out. I do enjoy making money — that part is fun.

Patreon, Then Substack: Pricing Content From Day 1

guest—DAVE GERHARDT—Founder Exit Five:
I had a pretty big following. Then it was natural progression — "I have a lot of followers, I'm going to launch a paid version." The closest comparison now would be Substack. Back then it was Patreon. So $10 a month to join my Patreon. People today are like "Patreon? What the hell is a Patreon? Are you making porn?" Where now everyone has a Substack they charge people for. My vehicle was a podcast. I felt the signals in my inbox — every week, no joke, getting messages from founders, investors saying "work with me, do this thing, do consulting?" Took those all as signals: dude, there's something there. People want the thing I'm good at.

I had no expectations. I honestly thought "if I make a couple hundred bucks a month, that contributes to rent." Three to six months in I had a thousand members. That was $10,000 a month in recurring revenue from content. That broke my brain because forever in my mind, $100K in a year was the epitome of success. I was making six figures from a blog about marketing essentially.

That was while I was a CMO. One thing happened to my advantage unfortunately: COVID. We went remote, I was home every day. I built that while still CMO at Privy. I had product-market fit before I even decided to go full-time. I never felt like I was taking a risk. My career has just been stacking things on top of each other that make sense.

What's Coming Next After Exit Five

host—JORDAN CHENEVIER-TRUCHET:
Do you have an idea of what's coming next?

guest—DAVE GERHARDT—Founder Exit Five:
I don't. But I know I'm going to take a risk. I don't think I'm ever going to retire. There's no point. The older I've gotten — approaching 40 — I'm obviously much wiser. I always rolled my eyes at older people giving advice, but I'm 38. I am smarter than I was at 25. Fact. I've had life experience.

If you don't use it, you lose it. I'm not going to retire. I want to keep making stuff. I'm a little nervous right now because so much is changing. I listened to Marc Andreessen talk to Lenny — everything, we're so screwed, AI is going to eat all the jobs. Then I'm like, should I be in AI? Should I build an AI? Then I'm using Claude every day and I'm like, tell me Claude wouldn't just replace every SaaS app in the world.

I'm bullish on real life. I'm going to own a car wash. I don't know what the idea is yet. The marketer and entrepreneur in me — I love writing, creating, communicating, public speaking. My bet would be: use my skills as a marketer to create another entrepreneurial journey after Exit Five.

Why Most B2B Communities Die

host—JORDAN CHENEVIER-TRUCHET:
Most B2B communities are like dead Slack groups. What did you do to make Exit Five work?

guest—DAVE GERHARDT—Founder Exit Five:
The dead Slack group thing — the real answer will be in five years. It is hard. A lot of communities go to zero. I think that's because they let them get over-riddled with spam and promotional stuff.

One way we solve for that early on is by charging money for it. People thought I was just an awful arrogant evil capitalist. But money is an amazing forcing function to get some level of commitment. We just did our event in Arizona. The two pieces of negative feedback we got were from two people who had free tickets. They were invited from a sponsor.

The Rules: Charge Money, Stay Niche, Police Comments

guest—DAVE GERHARDT—Founder Exit Five:
Number one, we charge from the beginning. Some level of commitment.

Number two, having a really focused niche. The niche of B2B marketing might not be small to you, but other communities apply to multiple areas — sales, marketing, revenue. From the beginning we've been exclusively for B2B marketers. You have to be in B2B.

Number three: I've written maybe 8,000 posts all-time and 40,000 comments. I'm doing it less now we have a team. But for the first five years, I'm in there every day — writing, commenting, policing comments. If Jordan writes a lazy post, I'm like "Hey Jordan, you should update your post — it's not very specific. What are you trying to get at?" Or "this is clearly you promoting your company." Founder mode. Things that don't scale.

Number four: we have a vibe and a way of doing things people relate to. I get messages all the time: "you're an authentic person, I really enjoyed meeting you in person."

The other thing: the value prop of Exit Five has evolved. It was just an online message board. We needed to evolve. Everyone churns for the same reason: "I didn't have enough time to log in." Like a gym membership you never cancel but never use. Two years ago I shifted from "online community in Slack" to a membership. Pay an annual membership fee and you get the discussion board, plus discounts, tickets to events, perks.

Communities mostly die because they're an acquisition channel, not the product. If a vendor owns it and the goal is nurturing people into a SaaS subscription, engagement drops. Every time. Exit Five isn't owned by a vendor. The community IS the product.

Storytelling as a Competitive Skill

host—JORDAN CHENEVIER-TRUCHET:
I see a lot of marketers who aren't good at storytelling, branding, communicating online. With AI, storytelling becomes the differentiator. Advice for them?

guest—DAVE GERHARDT—Founder Exit Five:
It's hard. It's almost like a friend who just has a way. The "je ne sais quoi" — that one thing you can't explain but it's there.

It's like that friend who's just kind of funny. Never annoying. Always down to hang out. Good time in any scenario. You can't go to ChatGPT or Claude and study a storytelling structure and just print it out and have it. I don't think our brains absorb communication that way.

Everyone overused the word "taste" a year ago when AI came out. There's something to it. It's not learning the storytelling framework. It's being curious, observing the world, having opinions about what makes something good, being able to make your version of it.

The people that are really good at marketing have a wide range of interests outside of work. "I like this artist, this musician, this chef, I like how this person does a landing page." Especially in B2B: we're always like "let's go see how Stripe does their website. Let's see how Shopify does their website." Where I'm like "shoot, check this out — this lady down the street has an amazing sandwich shop here in Vermont and her website is awesome."

The confidence to not just copy what's been done in your industry. Have a strong opinion: "yeah, it's okay, nobody writes this way, but we're going to write this way because the goal is to get attention."

Writing With Claude (and Why It Takes Two Hours)

host—JORDAN CHENEVIER-TRUCHET:
It's funny you mention taste. I just wrote a newsletter yesterday called "AI didn't kill creativity. It just proved you didn't have any."

guest—DAVE GERHARDT—Founder Exit Five:
I have a newsletter — exit5.com/newsletter. I use Claude to write it. Guess how long it takes?

host—JORDAN CHENEVIER-TRUCHET:
A long time, I imagine.

guest—DAVE GERHARDT—Founder Exit Five:
Two hours plus to write a really good one.

I'm the subject matter expert. So it's not "you used AI to write your newsletter." I didn't just give it to AI and one-shot it. It's: "I have this idea. I recorded a podcast with Jordan. Here's the transcript. Here's two other things I've written that relate to this. Here's an example of what good looks like. Go through this, organize it, then come back and ask me questions."

That's where Claude is amazing — like having a creative agency partner on the side. "Cool, got it Dave. Do you want the intro to be like this, this, or this?" I'm like "actually, fourth option — a personal story." Instead of writing start to finish, I'm piecing each section together.

The way I write: I never write start to finish. I have this thought here, "where does it go? That's a paragraph. That's an intro. Let me punch that up." I use Claude as my editing/research/writing assistant. The output is amazing pieces I put together.

It's certainly not saving me time. It's making the output better. As a person who likes to think I'm a writer and creator, the way I use Claude is awesome but it's not saving me time. It's making the output better.

Personal Brand vs Company Brand

host—JORDAN CHENEVIER-TRUCHET:
If you're a marketing leader inside a company — you don't own the company. How do you manage the tension between writing for the company versus writing for you?

guest—DAVE GERHARDT—Founder Exit Five:
It's easy for me — I run the business, I own it, no investors. But say you're VP of Marketing at a cybersecurity company. You can't write the same way I can. Get direct with the subject matter expert. You can use AI to get more effective content because in 15 years of marketing, the whole game has been "who are the experts inside our company?" Often developers, engineers, product people. Marketing has always needed to get them for interviews.

Now it's more possible because you can interview them, get the transcript, prompt with AI, get closer to being a subject matter expert than ever.

On personal: people buy from people. If you make a bunch of generic content with AI, it won't stand out. The opportunity is sprinkling in personality and being you. Being you for the company. That might be a false dichotomy. Every business is in the business of social media now because that's how the world gets information. We follow people. Look at LinkedIn engagement: a company page versus a personal page. That's the proof.

Metrics That Matter: Revenue, Churn, Engagement

host—JORDAN CHENEVIER-TRUCHET:
Do you look at data and dashboards for Exit Five?

guest—DAVE GERHARDT—Founder Exit Five:
I do. I don't obsess as much as I used to because I hired a CEO. That was me spending money to make it so I don't have to. I can call Dan or text him and ask "where are we at?"

I'm a competitive person — this is fun, almost escapism. I get to play this game of business on the internet.

The main things: Revenue. Money. Turns out you have to make money. Goal: make more than you spend. Quarterly P&L meetings. Indicators on those: growth in membership — are we getting new members? Churn — losing more than we're gaining? Churn is a killer. "Wait, we're spending all this effort to grow, but we're losing 60% of it. We should fix that."

Growing our email list. List growth, engagement, opens, clicks, replies, unsubscribe rate. NPS across all our products — are people happy with the thing they're paying for? Would they recommend it?

In B2B inside a company, we were told MQLs are bullshit, content downloads don't matter. But I run a media company — those ARE our metrics. The size and engagement of our list matters. We had an unsubscribe issue — we cut all our newsletter advertising because the list was growing but we were churning out more than we were gaining.

F*ck Attribution

host—JORDAN CHENEVIER-TRUCHET:
Do you work on attribution?

guest—DAVE GERHARDT—Founder Exit Five:
Fuck no.

It's so simple — that's why it's easy in our business. Attribution gets complicated when you have a big team, big budgets, spending across many channels, justifying ROI. We don't.

We do get attribution. We ask every person joining our email list or community: "how did you hear about us?" It's amazing because they tell us. "Dave's LinkedIn post. I saw an ad. I listened to the podcast. I subscribed to the newsletter." We export them a couple times a year. Honestly, we don't learn that much. It reaffirms what we're doing is working.

Newsletter drives a bunch of members. Inside a real company, you'd want to be like "this month seven people bought because of our podcast, so we should do more episodes." It just doesn't fucking work like that.

So much of this comes from over-metricing marketing. It's insane. I once interviewed a CMO of a billion-dollar company who said "yeah, we don't really measure anything." Tongue in cheek, but at some point you've seen the whole game and you just know — if you do the right things, people will show up. It's when we get caught in the middle, having to micro-measure every channel, that it gets crazy.

host—JORDAN CHENEVIER-TRUCHET:
So you don't think about the ROI of content.

guest—DAVE GERHARDT—Founder Exit Five:
Hell no. Because I know it. If we put out good content people like, they will join our newsletter, they will be on our podcasts, they make us look good, the brand looks good. It's a game of: if people think we're legit, they will. They're not morons. We need to keep showing up in the places they're at.

Closing

host—JORDAN CHENEVIER-TRUCHET:
Where should we send people?

guest—DAVE GERHARDT—Founder Exit Five:
exit5.com is good. Also I'm on LinkedIn — Dave Gerhardt. Connect with me, send a message.

host—JORDAN CHENEVIER-TRUCHET:
Thanks Dave.

guest—DAVE GERHARDT—Founder Exit Five:
Thanks for having me.

Introduction

host—JORDAN CHENEVIER-TRUCHET:
Hey, Eli!

guest—ELI SCHWARTZ—Author Product-Led SEO:
Bonjour Jordan, good to be here.

host—JORDAN CHENEVIER-TRUCHET:
Thank you for being on the pod. For people who don't know you, you're famous for having written a bestseller in the SEO space called Product-Led SEO. What is Product-Led SEO?

What is Product-Led SEO?

guest—ELI SCHWARTZ—Author Product-Led SEO:
The story is: I was working at SurveyMonkey — my last full-time job — and I started consulting on the side. A large brand reached out to potentially hire me. I gave them an offer and they asked: "So what are you going to do for us? Write content? Do keyword research on SEMrush or Ahrefs? Recommend links to build?" I said yes to all three.

Then he asked: "Why wouldn't I hire someone in the Philippines, Pakistan, Bangladesh, or anyone on Fiverr to do that?" I knew the answer was: I'm better. You reached out. You wanted to hire me. If you thought you could do it for $50 on Fiverr, you would have.

I didn't have the concrete answer at the time, and I didn't sign that client. But it forced me to think back on how I did things at SurveyMonkey. The way I did things was not just creating keyword-driven content and building links to it. I thought about what the user was searching for and built that asset. I needed a name for it — Product-Led SEO became the name.

What it really is: when someone is searching, they're not searching for something that matches a keyword. If someone searches "best mobile phone for me," they're not looking for a piece of content that says "best mobile phone for me" 60 times. They want the answer. They want a product. Product-Led SEO is building the thing — a piece of content, a video, an experience — that satisfies the answer, not just matches a keyword.

The Product Approach: TripAdvisor and Amazon

guest—ELI SCHWARTZ—Author Product-Led SEO:
TripAdvisor is a great example. When someone searches "review of the Ritz Carlton Paris," they're not looking for content that has the phrase "review of Ritz Carlton Paris" 60 times. They want the review. TripAdvisor's approach back then wasn't to write blog post after blog post. They built a website that satisfies any combination — Paris, New York, Ritz Carlton, Hilton. That's the product.

Same with Amazon. They didn't approach e-commerce the way late-90s competitors did — "let's make a piece of content about what a cell phone is." Amazon put all the effort into building the architecture and structure of great SEO. There are now millions of things sold on Amazon that were never dreamed about when Amazon was created — but they don't need to build a new Amazon for each. They extend the architecture. Even better, they took amazon.com and extended it to amazon.fr without needing a new SEO strategy for France.

host—JORDAN CHENEVIER-TRUCHET:
Would you say "product" is an umbrella word that includes content, reviews, calculators, etc.?

guest—ELI SCHWARTZ—Author Product-Led SEO:
It includes the thing the user is looking for. On a media site, content is the product. On an e-commerce site, the e-commerce landing page is the product. On a research page, it's the ideas. Product-Led SEO is really better defined by what it is not: content-led SEO — keyword research, create something that matches exactly what the user searches, call it a day.

Rewriting the Book for the AI Era

host—JORDAN CHENEVIER-TRUCHET:
Between when you wrote the book and today, a big thing happened: LLMs, AI, AI Overviews on Google. What would you rewrite today?

guest—ELI SCHWARTZ—Author Product-Led SEO:
A lot of my book was intended never to be obsolete. I wrote the whole book myself — I didn't hire a ghostwriter. Every time I used "Google," I added "Bing," just in case Bing would overtake Google one day. I didn't talk about a lot of tactics, again because I didn't want it to be obsolete.

Much of the book applies today. The big thing I'd change: I'd talk about AI being an answer itself. AI Overviews, Gemini, ChatGPT — they become that first-layer product. If a user is just looking for an answer, they no longer need a website that has done SEO.

The other part: I'm excited my vision for Product-Led SEO is now coming into fruition. The shady tactics — buying 80 million backlinks, spamming keywords — don't work as much anymore. What's left is essentially Product-Led SEO. The only thing I'd add: help the reader understand that experience also happens on an LLM, and you can't recreate it if it's happening there.

Will LLM Companies Need SEO?

host—JORDAN CHENEVIER-TRUCHET:
Would you advise these big LLM corporations to invest in SEO?

guest—ELI SCHWARTZ—Author Product-Led SEO:
I believe they already are. Go on LinkedIn — OpenAI, DeepSeek, Perplexity, Anthropic all have growth marketing teams and people with SEO titles.

Contrary to what everyone thinks based on social media — "Google's dead, everyone will use LLMs only" — I don't think that's close to the case. Those companies know that to reach 80–90% of the online world, they have to do it through Google. They don't have the liberty to say "we'll just find them on our own platform."

Given the antitrust suit in the US, Google maintained its monopoly. Google will dominate consumer search until something completely shifts. Not just because they're good — they have distribution power: Android, Workspace, Chrome. People say "kids aren't using Google anymore." In the US, many kids get their first device from school: a Chromebook. Google's Trojan horse is "here's your new computer, it can only use Chrome." Those kids may not search as much as older generations, but when they do, they use Google.

So those LLM companies know the way to reach online users is through Google and typical SEO.

The Search Pie Is Getting Bigger

host—JORDAN CHENEVIER-TRUCHET:
People see this as a shift from search to LLM, but actually the cake is getting bigger. People still use Google — they use Google plus OpenAI plus other LLMs. So Google may be losing market share, but not in absolute query volume.

guest—ELI SCHWARTZ—Author Product-Led SEO:
Exactly. There are many queries Google wasn't a good solution for. Four years before ChatGPT, everyone said Google was dead because of TikTok. TikTok satisfied an area of search Google couldn't — video restaurant reviews, trip plans.

Now LLMs do things Google couldn't. I was using Gemini to figure out why a door in my house wasn't working. I explained the problem, took a picture of the handle, Gemini gave me an answer. I said "that doesn't make sense," took another picture. I never could have done that on Google. I would have tried to find a forum post from someone with the same issue and copy their solution. This is a new area of search. By the end I'll probably need to buy a part — that's when I'll go to Google. That becomes a mid- or bottom-of-funnel search. But the top of funnel didn't even exist before.

Where AI Wins, Where SEO Wins

host—JORDAN CHENEVIER-TRUCHET:
You've said AI owns top-of-funnel and SEO wins mid-funnel and bottom-funnel — even with AI Overviews expanding into mid-funnel. Where do you draw the line?

guest—ELI SCHWARTZ—Author Product-Led SEO:
First: I'm not sure GEO is a real, separate investment. I'd continue to invest in SEO. Many big websites are going to get themselves in trouble because they think AI is the only future, and they'll make basic SEO mistakes — subdomains vs. subdirectories, technical SEO that should be structured at the start. Once you have traffic you don't need much technical SEO, but you definitely need it when setting up.

Top-of-funnel example: someone wonders if they have a health issue, like anxiety. That belongs to AI Overviews or an LLM. They describe symptoms, the LLM says "this sounds like anxiety." Many blog posts cover that, but Google's AI Overview takes the top of the page and gives the answer.

I worked with a mental health company. Their solution was therapy, for anxiety. There are many ways to address anxiety: meditate, listen to music, take drugs, go to therapy. But this company only offered therapy. So bottom-of-funnel was "therapy" or location-based queries. The super top-of-funnel — "anxiety" — doesn't exist for them anymore. AI Overviews handle that. Now AI tells the user "people with feelings like yours could meditate, listen to music, or try therapy." That brings them to mid-funnel: "therapy for anxiety." That's the only place this company can address.

What I recommended: focus on this very specific mid-funnel. The top will always belong to AI. The mid-funnel — therapy as the solution — is the only people they can address.

Citations Are the 2026 Version of Backlinks

host—JORDAN CHENEVIER-TRUCHET:
Lots of people read tactical advice — "change your titles to questions, do these tactics for GEO." But the data shows the top SEO results are 80% the top LLM results. So you should still invest in SEO even to appear on LLMs. Agree?

guest—ELI SCHWARTZ—Author Product-Led SEO:
Yes. I talk to so many companies that want me to recommend them for GEO/AEO. I ask "what's your silver bullet?" Their answer fits two buckets:

• Content directly matching what the user prompts. If they directly match, they show up. But we don't even know what the prompts are if they're long-tail. Their content gets cited but not necessarily recommended.
• Citations. Citations are the 2026 version of backlinks — and I don't think backlinks have mattered that much in the last couple of years. Google got smart about which backlinks were manipulated. Citations are the same. If the whole GEO play is getting citations, is that really new?

Beyond LLMs, AI is trying to think like a human. You can spam citations all you want, but if AI thinks like a human, it also thinks "which of these citations matter?" Most brands that get cited are cited because they're brands. Rather than figure out which button to push, build a brand users actually like. That's what the engines are looking for.

I asked John Mueller in Zurich a couple of months ago whether Google uses any data from Google Maps and Street View. He said of course they do. So if you claim an empty field is your headquarters, that picture feeds into Google search. If you have a restaurant and Street View goes by, Google Maps tells you if a store is busy. If I were an LLM with access to that data, I'd use it to recommend the pizza shop that's always busy at lunch. You can build all the citations you want for "best pizza in New York," but if Google knows no one goes there, it doesn't matter.

Basic SEO is what you need to show up in AI. There are nuances — if you are the best pizza and don't show up for "best pizza," treat it like any SEO problem. But don't run a whole "let's make sure everyone knows we're the best pizza" exercise if you already are.

The Five Stages of Awareness vs ToFu/MoFu/BoFu

host—JORDAN CHENEVIER-TRUCHET:
Many websites talk about a keyword without really understanding who the user is in the funnel. Are you familiar with the Five Stages of Awareness framework?

guest—ELI SCHWARTZ—Author Product-Led SEO:
Yes.

host—JORDAN CHENEVIER-TRUCHET:
Instead of ToFu/MoFu/BoFu, with the Five Stages of Awareness you focus on Problem Aware. Going back to your health example: Problem Aware is "I type my symptoms." I don't know any solutions, products, or specifics — I just know my problem. Of course I go to an LLM, because everything is foggy and the LLM helps me clarify. But when I know exactly what I need — Solution Aware, Product Aware, Most Aware — I don't need to discuss with a chatbot. I just go to Google to get my solution.

guest—ELI SCHWARTZ—Author Product-Led SEO:
Exactly. That's what many websites used to do with SEO. I had this experience with one of my first consulting clients, Mixpanel. We did all the SEO stuff — focused on keywords, built content. But none of it converted. Mixpanel is expensive and very sticky — you don't buy an analytics tool and rip it out a month later. People landing on the content were very early in the journey. Expecting them to check out and purchase right away was wrong. The right way: what do we do at this very high point of the funnel to bring them to the next part? Maybe retarget. Maybe get them to join a webinar. Not a big purchase.

A lot of SEO is "let's say you're Salesforce, you want to target the word CRM." Yes Salesforce is a CRM, but is that the right conversion keyword? Or is there another way of doing SEO that makes sense for the place in the journey the user is?

How to Vet GEO Agencies

host—JORDAN CHENEVIER-TRUCHET:
Every agency is selling GEO/AEO. What's one question a CMO should ask to figure out if these agencies actually know what they're doing?

guest—ELI SCHWARTZ—Author Product-Led SEO:
Start with the assumption that they probably don't. Agencies only selling AIO/AEO/GEO probably don't know what they're doing. If they have a track record of successful clients in SEO and now show how they've adapted SEO to AI, that's more trustworthy.

Same with tools: look at the people who made the tool. Do they have a track record of helping companies acquire customers from digital marketing? If they don't — if they just "found a niche, everything has switched, let's make money while people don't know what's going on" — avoid.

The basics of AI visibility are the same as SEO. So that's where you'd quiz them: can they help with traditional SEO? Do they understand best practices? Do they have a good client base? Then you can trust what they say on AI.

It is the Wild West with AI. Everyone's making things up. I see data studies that completely conflict and they're both right because the data is too vast. You can have one sample that proves exactly what you want, but it might not be a big enough sample to prove the whole internet.

Convincing the CEO Not to Over-Invest in GEO

host—JORDAN CHENEVIER-TRUCHET:
Let's say I'm a CMO. My CEO is pushing every day to invest in GEO. I agree with you but don't have the right arguments. How do I convince my CEO?

guest—ELI SCHWARTZ—Author Product-Led SEO:
You don't. I find it very difficult to convince people who really believe in something not to do it. So I never do outbound sales — I don't approach someone and say "your SEO sucks, hire me." I'm much more successful when they say "I have an SEO problem, can we work together?"

Same logic. If the CEO really believes GEO is the future, fighting creates friction. What I'd do: do it. Focus 10% of effort on GEO and AI visibility, focus 90% on typical SEO. In six months, if you're right that GEO doesn't matter, the CEO will say "this seems to be a scam, let's drop it." You can say "I tried to tell you, which is why I didn't put all my eggs in this basket." Now you deserve a bonus. You don't want to be the person who says "I tried, you wouldn't listen."

When I coach SEO employees with a manager insisting on a ridiculous keyword: come up with a plan, allocate a small budget, do some token effort. Spend 90% on the right things. In six months, when you don't rank on the dumb keyword, you can say "look at all the things I tried — and look at these other wins." At least you have something to point to.

Budget Allocation: $1M to Spend

host—JORDAN CHENEVIER-TRUCHET:
You have $1M as a CMO and need to allocate between SEO and alternatives — content, brand, community. Walk me through the decision tree.

guest—ELI SCHWARTZ—Author Product-Led SEO:
I used to ask this in SEO interviews. "You have $1M to spend before end of Q1. Three days left. How do you spend it?" I was looking for creativity. My favorite answer: take the $1M up in a helicopter and drop it. Everyone will talk about a million dollars falling out of the sky.

To your question: I would not spend it on SEO. First, it's very hard to spend $1M in SEO properly. Second, if you don't already have a brand, do brand marketing first. Most companies should do paid before SEO — it's quick feedback on what works, on product-market fit, on what I call search-market fit. Are people even searching for this? If not, your SEO won't work.

host—JORDAN CHENEVIER-TRUCHET:
When you say paid, you mean search ads, not social ads?

guest—ELI SCHWARTZ—Author Product-Led SEO:
Search ads, yes. I actually like social ads too — you can target users and figure out product-market fit. I'd do social ads first, then search ads, then SEO.

The biggest bang for buck for any startup with $1M is brand. Hire PR agencies. Get the name out. That trickles down into everything else — people search for you, social works, people sign up for emails. One of the biggest problems for startups is they raise money but no one's looking for the solution. Brand is what gets the ball rolling on every other channel.

Should SEO Sit in Product or in Marketing?

host—JORDAN CHENEVIER-TRUCHET:
Going back to Product-Led SEO — should SEO be in the product team?

guest—ELI SCHWARTZ—Author Product-Led SEO:
Ideally yes, but that's never a decision the SEO team gets to make. The reason: the levers for getting things done are better moved within the product team.

On a marketing team, you write tickets that go to product, then engineering, then back to marketing colleagues to design and write. Everyone fills out parts of what product does.

On a product team, you and your colleagues all report to the same boss and work on the structural assets. At a marketing team, marketers may not even know the engineers working on their requests. On product, you have engineering resources directly.

Example: making a CMS decision. At larger companies, marketing has little influence on that — it's product. Or homepage structure. Marketing can say how it kind of looks, but the assets feeding into it — hard for SEO to influence.

My most successful SEO engagement was Tinder. I worked directly for a VP of Growth. We had engineers, marketers, designers in our growth pod. Everything got done because nothing was walled off. Two years of work, and the impact came because I was in the product team.

host—JORDAN CHENEVIER-TRUCHET:
So what's the best org structure?

guest—ELI SCHWARTZ—Author Product-Led SEO:
SEO would be a product manager whose responsibility is SEO. They have engineering, marketing, design resources reporting in. Hub and spoke, with the PM at the center. If you're SEO inside marketing, you become a spoke and lose control.

Technical SEO and Link Building Are Overrated

host—JORDAN CHENEVIER-TRUCHET:
I heard you say link building is overrated and technical SEO is over-indexed. What do you mean?

guest—ELI SCHWARTZ—Author Product-Led SEO:
When companies look for an SEO consultant or agency, they often start with an SEO audit. The audit comes back with technical feedback — 301 redirects, 302s, 404s. If you implement everything on the audit, you're unlikely to see much growth. That's why I think it's over-indexed.

It's like going to the doctor and being told to run 10 miles a week. Will it make you live longer? Maybe, maybe not. Yes, it's good not to have 404 pages. Will you meaningfully make more revenue? Most honest consultants would have a hard time saying yes.

Technical SEO is going to be under-indexed and skipped in AI because everyone's like "we just want AI visibility and citations." For bigger companies, structure will matter and they'll skip it.

On link building: there was a time it really mattered — raw PageRank. Google looked at PageRank as critical and didn't have great filters. For many years now they filter that out. I had a backlink from WhiteHouse.gov when I joined SurveyMonkey. The Obama White House did a partnership, presented a survey ceremony with Michelle Obama. The blog post linked to a 404 on SurveyMonkey. You don't get to ask the White House to update the link — so I created a page that made the 404 go away, then linked from that page across the site. All those pages started ranking higher. That was 2012.

Then in 2018 SurveyMonkey got more backlinks from the Trump White House — they were scraping our research. None of those did anything. Why? Google got smarter. The White House is authoritative on government and the economy, not on customer feedback surveys. So Google says "this is not a relevant link, we won't pass authority."

That was 2018 — eight years ago. They've gotten even smarter. They look at a Forbes link and say "this is paid placement, not a relevant link." That's an algorithmic challenge they've gotten closer to solving. So just getting a high-DA link isn't something I'd invest in. Same with citations: there will be good ones and bad ones, and just getting them because they're authoritative isn't a good investment.

The Future of SEO

host—JORDAN CHENEVIER-TRUCHET:
How do you see the future of SEO?

guest—ELI SCHWARTZ—Author Product-Led SEO:
SEO never disappears. We're going into a world where search will always exist. I made a prediction recently that Meta is about to launch its own search engine — their LLM already exists at meta.ai. It will work by pulling Meta data — Facebook pages, Instagram, WhatsApp Business. They've made big AI acquisitions and have data centers. Real player.

host—JORDAN CHENEVIER-TRUCHET:
They actually announced an acquisition today.

guest—ELI SCHWARTZ—Author Product-Led SEO:
Yes, multiple. The world of LLMs will be dominated by companies that were already huge. OpenAI raised a lot, but beyond OpenAI and Anthropic, big tech wins. I don't see a startup just showing up and challenging Google or Meta.

SEO is always this request for discovery. It's the opposite side of the coin from paid marketing. From newspaper ads to TikTok ads. Search 15 years ago was "go on your computer and type into Google." Next year it might be glasses made by Meta or Google — you look at something and get an answer. That's still search.

The job of SEO is to optimize for that experience. You don't want to leave it to chance. Meta requires X to show up in glasses results — the SEO team makes sure your asset shows up. You don't say "we exist, hope it figures itself out."

The Most Contrarian Take

host—JORDAN CHENEVIER-TRUCHET:
One contrarian take that you believe almost nobody in the industry agrees with you on?

guest—ELI SCHWARTZ—Author Product-Led SEO:
I really don't think anybody should be investing significantly in AI as a separate thing. Invest significantly in search visibility. The way GEO/AEO is framed today is as if it's something separate. It's like 20 years ago when "mobile SEO" became a thing — as if you had an SEO person and a mobile SEO person. Or international SEO — same trap. They're all one and the same.

AI visibility is just the 2026/2027 version of SEO. Make sure your existing agency knows how to think about SEO of today, which has an AI visibility element. Don't break it off and call it a new thing — that creates a budget category you don't have to spend in. Same as "we can't afford mobile SEO, hopefully people find us on desktop." Wrong approach.

I've invested in multiple AI visibility tools because I'm curious where they go and I want to attach to growth if I'm wrong. But it shouldn't be a separate investment.

Closing

host—JORDAN CHENEVIER-TRUCHET:
There's money to be made fueling that narrative. Fair enough. Thank you, Eli. For people who want to follow you, you've got a newsletter and you post on LinkedIn. Your book Product-Led SEO is still a bestseller — link in the description.

guest—ELI SCHWARTZ—Author Product-Led SEO:
LinkedIn is great. And please read my newsletter. I've been doing it for three years. When I started, no one read it — I had no subscribers. I just wrote to the world. Now I have almost 14,000. Whenever someone asks how I got there, I just kept writing. Honored if anyone reads it.

host—JORDAN CHENEVIER-TRUCHET:
I'll put the link in the description. Thank you, Eli, and all the best.

guest—ELI SCHWARTZ—Author Product-Led SEO:
Thanks for having me.

Introduction

host—JORDAN CHENEVIER-TRUCHET:
Welcome, Adrian.

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
Merci. Thank you.

host—JORDAN CHENEVIER-TRUCHET:
So you're here for the Momentum Tour?

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
Yes.

host—JORDAN CHENEVIER-TRUCHET:
Can you just speak about that tour? What is it?

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
Absolutely. At Webflow, we saw an amazing opportunity across the world with incredible customers. We've seen so much growth across the European business, with customers like Spotify, monday.com, and Anytime Fitness.

We really wanted to go on the road with our customers and partners to talk about how the world is changing and what we're seeing in the web space. It was an opportunity to learn from them while also sharing what we’re seeing across our 300,000 customers so we can meet this moment together, because so much is changing so quickly.

Recording a First English Podcast

host—JORDAN CHENEVIER-TRUCHET:
For my audience, this is the first time they’ve heard me speak English, so apologies if I make mistakes. This is my first time recording a podcast in English, so I’m glad to do it with you.

Today, Adrian, you’re the CRO of Webflow and you’ve been there for about nine months. I’d like to understand your onboarding and your first 90 days. At this level of responsibility, those first months are critical to making an impact. How did you approach them?

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
When I joined Webflow, the world was already changing quickly in the web industry, but I never imagined how fast AI would reshape discovery and search.

In my first 90 days, I focused on understanding three perspectives:

• What our customers say about us
• What our partners say about us
• What our employees say about us

I wanted to understand our place in the market, our impact, and where the future was heading.

Since I oversee sales, marketing, support, and customer success, I also spent significant time understanding where revenue comes from, how it evolves, and how we could build a vision for the future alongside our customers, agencies, and partners.

Biggest Surprise After Joining Webflow

host—JORDAN CHENEVIER-TRUCHET:
What surprised you the most when you arrived?

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
The biggest surprise was how quickly people’s behavior around information consumption changed.

Back in early 2025, AI-assisted search existed, but it wasn’t yet materially different for most people. Depending on where you lived, AI web search was still limited.

What surprised me was how quickly the world changed in the way it consumes information online.

There are now major conversations happening around:

• What is the role of the website in an AI-first world?
• What happens in six months, one year, or two years?
• How does discovery change when AI answers questions directly?

On a positive note, one of the most rewarding surprises has been seeing what agencies and customers build. The creativity, visual quality, and immersive experiences they create are incredible.

Transitioning from Salesforce and Tableau to Webflow

host—JORDAN CHENEVIER-TRUCHET:
You come from Salesforce and Tableau, which are more traditional B2B environments. Webflow has a much stronger design and creative culture. What changed in your role?

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
The mission at Webflow is to bring developer superpowers to everyone.

That means helping customers make creativity their competitive edge.

The biggest shift is prioritization. At Webflow, creativity is central:

• How do we help customers express creativity?
• How do we create beautiful experiences?
• How do we maintain a design-first culture internally?

Because we’re a design-first company, there’s a higher bar for delivering excellent experiences.

For example, relaunching Webflow.com brought pressure to meet those expectations—and I enjoy that challenge.

Creativity Inside Teams

host—JORDAN CHENEVIER-TRUCHET:
Do you have more creative people in your teams compared to previous companies?

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
I wouldn’t necessarily say more creative people.

The difference is that creativity becomes a top priority.

Creativity shows up in many forms:

• Design
• Problem-solving
• Strategy
• Innovation

At Webflow, it’s simply much more front-of-mind.

host—JORDAN CHENEVIER-TRUCHET:
How do you encourage that creativity?

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
I push people to be bold.

I often tell my team:

“I want you to make me uncomfortable with your idea.”

If an idea makes me uncomfortable, it usually means we’re pushing toward the edge of what’s possible.

We’re operating in a rapidly evolving world, and creativity matters not only for humans anymore—but for AI as well.

We now design for:

• Humans
• AI systems consuming websites

That requires a different philosophy.

Managing Sales, Marketing, Success, and Support Together

host—JORDAN CHENEVIER-TRUCHET:
As CRO, you oversee sales, marketing, support, and success. How do you balance optimizing individual functions while keeping the revenue engine aligned?

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
I think about this as rows and columns.

Imagine a spreadsheet:

• Rows represent products
• Columns represent functions like sales or marketing

Each row and column has an owner.

The key is ensuring alignment across functions so nobody works in isolation.

I often tell the team:

“We win together, and we never lose alone.”

That means:

• No initiative should exist without cross-functional ownership
• Teams should have shared accountability
• Collaboration must be structurally built into the organization

Where Revenue Engines Usually Break

host—JORDAN CHENEVIER-TRUCHET:
Where do revenue engines usually fail?

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
Most businesses spend too much time diagnosing the past.

Instead of traditional business reviews focused only on what happened, I run a process called “Plan to Make Plan.”

Rather than asking:

• “What happened last month?”

We ask:

• “What did we learn?”
• “What will we do differently?”
• “Who will partner with us to execute?”

Revenue engines become harder to manage when:

• Multiple lead sources exist
• Product-led growth combines with sales-led growth
• Funnel complexity increases

The goal is to diagnose quickly and adapt before momentum is lost.

Team Operating Rhythm

host—JORDAN CHENEVIER-TRUCHET:
What routines do you use to manage this?

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
At the start of every month, every leader runs a “Plan to Make Plan” review.

It includes cross-functional partners.

We also run:

• Weekly forecast calls
• Cross-functional revenue reviews
• Monthly alignment conversations

The rhythm matters because consistency creates accountability.

Dashboard and Metrics

host—JORDAN CHENEVIER-TRUCHET:
What data do you monitor daily?

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
I built my own dashboard called “State of the Union.”

I monitor:

• Revenue by source
• Geographic performance
• Pipeline creation
• Inbound vs outbound performance
• Partner-generated revenue
• Product-led vs sales-led motion

I also log into Webflow daily to understand:

• Traffic trends
• Homepage performance
• Pricing page engagement
• Product page engagement
• Referral sources

Staying Close to Customers

host—JORDAN CHENEVIER-TRUCHET:
How deeply do you go into customer conversations?

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
As deep as necessary.

I strongly believe you don’t truly understand something until you talk to customers.

When I create a messaging hypothesis, I’ll test it directly with customers within 24 hours.

That prevents strategy from becoming overly academic.

The Momentum Tour has been valuable because conversations in Paris, London, and Dubai continuously reshape the messaging based on what customers are experiencing.

Why Smaller Events Matter

host—JORDAN CHENEVIER-TRUCHET:
I was surprised to see you attending smaller events. It feels unusual for someone in your role.

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
Smaller events often produce higher-quality conversations.

Right now, there’s no better place to understand change than by bringing leaders together and learning how they are adapting.

International Expansion and Local Presence

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
We saw an opportunity to create more success by bringing our teams closer to customers.

Being close to customers is how you:

• Learn faster
• Collaborate better
• Build trust

We’ve expanded teams across:

• London
• France
• Germany
• Nordic regions

Research we conducted showed that 91% of marketers identified their website as their most valuable marketing asset.

Because websites drive revenue, it’s important for us to support customers not just technologically—but strategically and locally.

Global Standards vs Local Innovation

host—JORDAN CHENEVIER-TRUCHET:
How do you draw the line between global and local teams?

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
I like the phrase:

“Global standards, local innovation.”

The idea is:

• Create consistent global frameworks
• Allow local teams to adapt to market realities

For example, data about AI trends may be global—but how you communicate that message should be shaped by local customer conversations.

AI Differences Across Countries

host—JORDAN CHENEVIER-TRUCHET:
Different countries are experiencing AI differently. France doesn’t yet have AI Overviews in Google.

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
Exactly.

Some countries are already testing AI-native search experiences, while others are earlier in adoption.

That’s why local market understanding matters.

You need to think not only about where the market is today, but where it will be:

• In 3 months
• In 6 months
• In 12 months

Support, Marketing, and Local Expertise

host—JORDAN CHENEVIER-TRUCHET:
Do you approach global/local structure differently across marketing, support, and success?

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
Support is global and operates 24/7.

AI has improved support efficiency because it allows faster multilingual responses.

Marketing is more centralized, but revenue ownership requires local alignment.

For example:

• A sales leader for Europe must have marketing alignment
• Customer success requires regional expertise
• Solutions architects often need local specialization

Customers want to speak with people who understand their market realities.

Partnerships as a Growth Lever

host—JORDAN CHENEVIER-TRUCHET:
Partnerships seem to work especially well at Webflow. How do you think about them?

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
Agencies are not just software implementers.

They are strategic partners.

In a rapidly changing market, agencies provide:

• Specialized expertise
• Creative thinking
• Strategic guidance
• Technical execution

As AI evolves, implementation becomes easier—but creativity and strategy become more valuable.

We’ve built partnership success into:

• Team incentives
• Organizational structure
• Executive prioritization

That’s why partnerships are a strategic investment.

AI’s Impact on the Web

host—JORDAN CHENEVIER-TRUCHET:
You sit on a lot of data. What do you see happening with AI and websites?

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
One thing is becoming clear:

Websites are no longer built for humans alone.

We’ve seen:

• A 125% increase in AI crawler activity
• A 6× increase in conversion from LLM-driven traffic

This suggests users arriving from AI are more informed and further down the funnel.

The challenge becomes:

• How do we structure content for AI understanding?
• How do we create differentiated human experiences?
• How do we adapt websites for both audiences?

AI Strategy Framework

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
We think about AI through two lenses:

1. Owning the Answer

This includes:

• Authority
• Measurement
• Content
• Technical setup

2. Converting the Traffic

This means creating immersive experiences once users arrive.

Websites need to feel:

• Relevant
• Personalized
• Interactive
• Visually engaging

Why Wikipedia Became Less Relevant

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
I realized recently that I don’t visit Wikipedia anymore.

The information still exists—but the experience hasn’t evolved.

AI provides answers faster.

This illustrates a broader truth:

Information alone is no longer enough.

Experience matters.

Are Websites Dead?

host—JORDAN CHENEVIER-TRUCHET:
There’s a debate that websites are dying because of zero-click content. What’s your view?

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
For websites to disappear entirely, you’d need a world where:

• AI no longer needs websites to learn
• Synthetic data replaces real web content
• Brands no longer need owned experiences

We’re far from that reality.

AI still depends on websites for understanding brands, products, and positioning.

What changes is the role of the website.

The website becomes:

• A structured source of truth for AI
• A branded experience for humans
• A dynamic environment for engagement

The Future of Marketing and Website Design

host—JORDAN CHENEVIER-TRUCHET:
Marketing is returning to first principles.

Clear messaging matters again.

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
I agree.

The future rewards:

• Clear communication
• Strong user experiences
• Practical messaging
• Better digital design

AI will force companies to become more intentional.

Adrian’s Biggest Challenge

host—JORDAN CHENEVIER-TRUCHET:
What is your biggest challenge today?

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
Helping customers adapt to rapid change.

The rules are changing quickly.

Customers need help understanding:

• How LLM changes affect discoverability
• How websites should evolve
• How to structure content for AI

My focus is helping customers navigate uncertainty while remaining competitive.

Closing

host—JORDAN CHENEVIER-TRUCHET:
Thank you very much, Adrian.

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
Thank you.

For anyone interested, we also created an AEO maturity model that benchmarks your website against trends we’re seeing across customers.

It provides practical recommendations and is freely available.

host—JORDAN CHENEVIER-TRUCHET:
I’ll include the link in the description.

guest—ADRIAN ROSENKRANZ—CRO at Webflow:
Perfect.

Outro

Speaker #2:
“Just do it. Don’t let your dreams be dreams. Yesterday you said tomorrow. So just do it. Make your dreams come true.”

‍