Human, Data, AI: Johan Abadie’s Blueprint for Modern Demand Generation

Johan Abadie runs marketing at SalesIntel, and is also his own buyer persona — a demand generation operator selling to demand generation operators. He makes the case that cheap contact data is the most expensive line item in your stack, gives the clearest working explanation of first, second and third party intent you are likely to hear, and argues the contrarian position that AI search makes paid search more important rather than less. Plus three specific places AI is already doing real work inside his team.

Show Notes

Johan Abadie runs marketing at SalesIntel, a pipeline generation platform competing directly with ZoomInfo. He's also, unusually, his own buyer persona — a demand gen operator selling to demand gen operators. That double vision runs through the whole conversation.

Cheap contacts are expensive

Johan's sharpest argument is about the hidden cost of bad data, and it applies well beyond his category.

"I think they are forgetting that cheap contacts are expensive. It's just expensive when it's cheap, because the connect rate is not as high, the bounce rate is a lot higher on your sequences."

The visible cost is the subscription. The invisible cost is an SDR team that looks busy and produces less — slower output disguised as higher activity.

He pairs that with a decay statistic worth sitting with: Gartner has found up to 70% of B2B contacts go obsolete within a year. Out of a million records, seven hundred thousand rot annually. Meanwhile most companies refresh their CRM once or twice a year, and build territories, account assignments and sequences on top of it.

His broader point lands harder in an AI era:

"Everyone is so hyped up about AI. If your data is not good, there is no AI tool that will make you perform well. The foundational part when it comes to AI is the quality of your data and the quality of your processes."

First, second and third party intent

Johan gives the clearest working explanation of intent data you're likely to hear.

The problem: you have 50,000 target accounts and a $10,000 budget. Advertising to all of them doesn't work. So you narrow.

Third party intent — providers like Bombora tracking topic-level interest across the web — tells you which accounts are showing a spike in topics adjacent to what you sell. Second party — G2, TrustRadius — tells you who is actively comparing in your category. You aim advertising at those accounts to drive them to your site. Once they arrive, that becomes first party intent, and the campaign changes entirely:

"Whenever someone goes on your website, then you can be super aggressive with these campaigns, because now they know you, so you're going to try to push them down the funnel."

His framing for why this matters at all:

"Try to sell me diapers. My kids are 15 and 16 and a half. Doesn't matter how good of a deal you're going to give me, I'm not going to buy diapers. I'm just not in market for it."

Bidding on your own brand

Asked about brand versus demand, Johan's answer is that you don't get to choose — do only one and "you are not going to be head of demand gen very long." But he singles out one non-negotiable.

Competitors bid on his brand name. If he doesn't defend it, the paid results above the organic ones belong to ZoomInfo and 6sense.

"No one is going to beat us on our brand. It will happen once in a while, but we should get 90-plus percent on the impression. That's unacceptable to lose."

Expensive, and table stakes.

Why AI makes paid search *more* important

This is Johan's most contrarian view. The consensus is that AI search erodes paid search. He argues the opposite.

Education is moving from your blog to ChatGPT and its peers. Buyers arrive already informed — which means the traffic that does reach your site is more transactional and more valuable.

"The education is shifting from your blog to ChatGPT and all of the LLMs, and now the traffic that you are going to get on your website based on this education is going to be more transactional. So in many ways, don't screw it up."

On optimizing for LLMs versus humans, he stays on the human side — structure your content properly, index it correctly, but bet on good original content winning.

He also flags a problem most people haven't noticed: you can't trust your own search results as a measure. LLMs personalize based on prior conversations.

"If I ask for providers of solutions, SalesIntel is always the first one for me with ChatGPT. I guarantee you it's not the case if you do that search."

He worries the echo chamber will harden over time, making it progressively harder for new entrants to break in.

Three levels of AI in a marketing team

Johan is specific and practical, which is rarer than it should be.

Content. He builds landing pages with Gemini — Google's own tool, on the theory that Google knows landing page optimization. He supplies the value prop, audience, tone examples and competitor pages, then edits.

"You don't become a content writer anymore. You're an editor."

The org implication is blunt: you probably still need one senior content person, but not three.

Market research. Deep research across tools, with a prompt discipline he insists on:

"Do not make things up. Do not extrapolate. Do not invent anything that is not a reference that has been cited."

Enablement. His favorite. Marketing produces case studies that sellers never find or use. Load them all into NotebookLM and a rep can ask for the supply chain case study that speaks to revenue growth — and get exactly that, including an audio version, without hallucination, because it's constrained to the source material.

"Case studies is absolutely the best content ever. It's the proxy — it's the person who is like you trying to do the same thing as you are doing."

The part that keeps him up at night

Johan used the analytics tool Julius.ai for an analysis of which keywords drove late-stage pipeline — pulling data out of Salesforce, HubSpot and Marketo into one view.

"It would have taken me two weeks to do that two years ago. It took me an hour and a half."

Which creates a new problem. He can now produce work that used to require specialists, and everyone wants more of it — but he never gets time to package it so anyone else can do it.

"It gives the illusion that you are the smartest guy in the room, but you're not."

Asked what worries him, it isn't the technology. It's building a team that can operate without him, when the tools change monthly.

"It's not the strongest, it's not the fastest, it's not the smartest that will survive — it's the most adaptable. But we are humans. I need a little time to get that done, and the market doesn't give you that time anymore."