How Ramp is turning its marketers into builders with AI3
3 Lessons From How Ramp Does Marketing
They put an actor from The Office in a glass box in NYC to file paper expenses for twelve hours and live-streamed the whole thing.
An employee got married inside the box.
They are sponsoring pro athletes. There’s billboards everywhere.
But there’s also a whole lot of scrappy, everyday marketing that you and me can do with no budget too.
Ramp might be building one of the best B2B brands of the last decade, and while the product rocks, I give a bunch of credit to the marketing team.
Spend management companies of this kind have been around forever. It’s essentially a corporate card. Expense reports. Not exactly the sexiest category in software. But Ramp has managed to make it feel different, and a big part of that is their marketing and brand.
I’ve been paying attention to how they operate for a while now. A few months ago we did a webinar on ABM with folks from Ramp, Snowflake, and Hightouch. Drew Pinta, who leads the growth data effort at Ramp, was on the panel, and he stood out to me as a sharp thinker. He’s not a marketer by title, but his role is basically being a strategic partner to the marketing team on everything from measurement to budget allocation — and I thought he’d be an interesting guest for a deeper conversation on the podcast about how Ramp is doing marketing today.
For today’s newsletter, I’ve got three big takeaways from my conversation with Drew that you can ponder while you’re sipping your morning coffee. Or creatine and electrolytes. Or green juice. Or tea. Whatever your thing is, I know these lessons will apply.
Here’s the full conversation with Drew up now on my YouTube channel if you want to watch btw.
1. The fastest way to kill creative marketing is to measure it like direct response
Drew said something I haven’t been able to stop thinking about: “The fastest way to kill creative marketing ideas is to try to measure them like direct response.” Put that on a bumper sticker! Maybe a long one, but still you get the point.
Too many teams force their marketing to fit the measurement system they already have. You run a brand stunt or sponsor a podcast, and then someone asks for the SQL count from that campaign. And when the numbers don’t show up in the MTA model, leadership loses confidence and the creative work gets cut.
The philosophy at Ramp: the measurement should meet the marketing, not the other way around.
An example: after the glass box stunt, they ran what Drew calls an “event study.” They looked at baseline marketing performance, forecasted what the next week would look like without the stunt, then compared it to what actually happened. Simple concept, no data science team required.
But the bigger unlock was using AI to scan thousands of Gong call transcripts. They had Claude look for every time a prospect mentioned how they heard about Ramp. What they found: LinkedIn was mentioned three times more than Meta. Their MTA model had said the exact opposite. That single insight led to substantial budget reallocation decisions.
You don’t need a big budget like Ramp or a fancy attribution stack to do this. You need call recordings, an AI tool, and the willingness to ask: what are customers actually telling us?
2. Split your budget 70/30: proven channels vs. experimental bets
One of the biggest mistakes I made in my career as a marketing leader is not having enough budget in the test and learn bucket.
This creates a challenge because you’re so short-term focused on hitting the goals today that when the plan grows next year and into the future, you don’t have enough new channels to help scale marketing.
Here’s a benchmark from Ramp: they use a simple 70/30 budget split. Seventy percent goes to core channels they’ve already proven out (Google, LinkedIn, Meta). Thirty percent goes to experimental bets they’re trying to scale for the future.
This sounds simple, but the part most teams get wrong is protecting that 30%. Drew was clear: you need to work this out with finance ahead of time. If your experimental budget gets judged on the same LTV-to-CAC metrics as your core channels, nobody will ever take a risk. You need a structure where leadership isn’t looking at your overall numbers and panicking because a new channel test tanked your blended efficiency.
They also have a channel maturity framework they run new channels through before they graduate to “core.” The gates include things like: Do we have solid optimization events set up? Is it showing up in our how-did-you-hear-about-us survey? Are prospects mentioning it on sales calls? They’ll run cheaper in-platform incrementality tests first (like Meta’s built-in holdout tests), and only move to more expensive geo-based incrementality tests once a channel has proven itself at the earlier stages.
The lesson: don’t build your annual plan on assumptions about unproven channels. And don’t kill experimental channels because they didn’t perform like your core ones in month one. This is how you think both short-term and long-term — by having budget and time allocated to develop new channels before you need them.
3. Ramp is turning its marketers into builders with AI
Everyone right now is talking about AI adoption in marketing, but Ramp made this a company-wide initiative.
Ramp’s CEO actually told the team: it’s okay to drop the ball on parts of your current job if it means learning to rebuild that work with AI. That’s not a vague AI pep talk. That’s permission to change what the job looks like (read this later from Ramp’s head of internal AI: We Built Every Employee at Ramp Their Own AI Coworker)
Drew said 80% of what he was doing nine months ago has now been automated. And somehow once again I feel completely inadequate btw…
Ramp’s marketers are using Claude Code to build their own dashboards, spin up go-to-market strategies for new verticals, and create tools that used to require a data science team.
The best example: a marketer named Alana runs Ramp’s vertical marketing. She spent a year and a half manually building out their construction vertical — doing the research, building landing pages, prepping sales materials. Instead of hiring three more people to do the same for three more verticals, she built what they call the “vertical machine.” It’s an agent that takes a new vertical as input, scans Gong calls and public research for pain points and keywords, spins up hundreds of SEO pages, and deploys a full go-to-market strategy with a click.
Drew’s bull case for marketers in the age of AI: AI gives you the median output, and in marketing, the median is death. Average ads don’t stand out. Average copy doesn’t convert. The human element — creativity, taste, knowing what’s worth building — that’s where the value is.
His bear case: 80% of what marketers do today is going away. You can reinvent yourself now, or you can wait for the market to force it on you.
So what’s it going to be?
– Dave
P.S. Where does your team stand on AI adoption right now? Does Ramp feel miles ahead of you, or closer than you think? Hit reply, I want the honest answer.
Also, I put the full conversation with Drew up on my YouTube channel right here.
