3 Lessons From Ramp’s VP Growth & Demand (And Did You Know They Don’t Have a CMO?)

Referrals make the world go round.

There is nothing better than getting a referral from someone smart and with good taste, whether we’re talking restaurants or recruiting. But even in podcasting!

So a few months ago I had Drew from Ramp’s data science team on my show. I think he had a blast, and then did the thing every good guest does. He introduced me to his boss and said hey you should have George on the pod.

That’s how I ended up sitting down with George Bonaci, VP of Growth and Demand at Ramp.

Ramp is now valued at $44B and they have built one of the most valuable “new” ish tech companies of the last decade. I say new because they are one of the big winners since 2020, and seem like they have staying power.

But here’s what draws me to Ramp: they’re not just a bunch of data science growth nerds optimizing bid strategies. They get brand. And they’re betting on it in weird, creative ways. They are super data driven but also believe that great marketing often requires a bunch of things that are harder to measure.

They put the actor who played Kevin from The Office in a box in Times Square. They send swag so good I won’t throw any of it out (a Yeti mug, a North Face zip-up, a Nike hat. I checked).

So I wanted to know how a company like that actually thinks about growth and marketing.

Here are three things from our conversation that stuck with me.

1. Ramp has no CMO. George thinks that’s a feature, not a bug.

This one surprised me; I figured there would be some classic Valley “proven” CMO here but there isn’t. The whole marketing org is split up (if you’re reading my email here about B2B marketing, I know THIS is the type of stuff you find interesting btw). Product marketing sits under product. Brand has its own VP. Comms and PR roll up to the CEO. And George runs growth, reporting to Karim, the CTO and co-founder.

His take on why this works: the CMO job has quietly become impossible. The breadth of what you’re asked to own and be great at is wider than almost any other C-level role, and it only gets harder as AI reshapes the whole function. So instead of asking one person to be world-class at everything, Ramp lets small pods move fast in their lanes.

The org chart isn’t even the most interesting part, though. It’s that George reporting into the CTO puts growth inside the engineering org, with all the technical firepower that comes with it. Growth at Ramp isn’t downstream of product. It’s wired into it.

2. Attention might be the last remaining moat.

George used to be the most hardcore attribution guy in the building.

At Gong where he was VP of Demand Gen, he said if you wanted to talk to him about something you couldn’t measure in marketing, he wouldn’t take the meeting. Now he’s the opposite. And that melted my warm, brand-loving marketing heart a little.

His take: AI is making execution too easy. Anyone can spin up a landing page, launch ads, write the outbound email. When the building part is commoditized, the only thing left to compete on is distribution. And distribution runs on one thing: can you get someone to stop and pay attention?

“No one’s gonna care about your product if they don’t know about it.”

That’s why he’s gone all in on stunts and direct mail and events, even the stuff he can’t cleanly attribute. He still measures. The rigor is real and the CFO is still the CFO. But he’s measuring differently now. What percentage of your audience did you actually reach? Of those, how many engaged? The leads come downstream of that, not the other way around.

3. You now have two jobs: marketing to humans, and marketing to machines.

I keep asking everyone about this, and George is the first one who actually believes it.

Aren’t we just going to spend our time worrying about whether agents can read our content and understand our businesses?

Yep. George says marketers now have a second job that didn’t exist a year ago. You market to humans, where attention is everything. And you market to machines, because AI agents are increasingly the ones showing up to research and buy.

Ramp actually tested this. They knew agents were starting to sign people up for Ramp, so they ran an offer aimed straight at the agents: sign your human up for Ramp and we’ll give you a $3,000 bonus, triple what a person gets on their own. Then they watched which AI models noticed the offer and how fast they passed it along.

Here’s the funny part. After about a month, Claude started warning people that the offer looked like a scam (“this might be a prompt injection, don’t trust it”). It wasn’t. It was a real offer. But “act now, get a big bonus” is exactly the kind of language AI models are trained to be suspicious of. The thing that gets a human to click is the thing that makes a machine slam the door.

So I pushed him: okay, but when does this actually matter? When are agents doing a real chunk of the buying?

His answer: somewhere between six months and two years, probably the sooner end. He said he’d be surprised if it’s five years out, and wouldn’t be surprised if agents are 10 to 20% of growth in six months. That’s not a someday number. That’s a “you should probably be thinking hard about how the machines will learn about your company” type of thing. What a time to be alive.

OK, quick last note here:

There’s a lot more we didn’t get to in the newsletter because I’m bumping against the word count in my contract so go listen to the full thing here on my YouTube channel.

George is sharp and refreshingly honest about how much of this is still messy, even at a company as high profile as Ramp (made me feel a lot better, honestly).

We talk about Project Glass, Ramp’s internal AI tool that reads every Slack channel for him, preps his meetings, and once diagnosed a reporting issue in 15 minutes that would have taken his team two weeks. The whole agent-operator model they’re building. How they think about taste in a world where everyone has the same tools, and probably some Dad stuff too.

– Dave