Most of what gets called personalization at scale in B2B marketing is a first-name token in an email, which isn't really personalization at all. It's a mail merge wearing a nicer jacket.
I learned the difference the hard way, building an account-based program from zero earlier in my career. There was no playbook, no existing pipeline, and no martech stack worth the name. What follows is what I actually did, what it returned, and what I'd change if I were starting today.
The problem
We were selling a high-consideration product to a finite set of accounts. The sales cycle was long, the deal sizes were large, and the buying committee was never one person. Spray-and-pray demand gen was useless in that context, because generic ads burned budget on people who would never buy, and the few good-fit accounts got the same treatment as everyone else.
The real problem wasn't reach, it was relevance. I needed the right accounts to feel like we'd built the experience for them specifically, and I had to do it without hiring a hundred people to handle it by hand.
The constraint
I had no team to throw at manual one-to-one campaigns, and no budget to waste testing into a void. So the system had to do two things at once: behave like bespoke outreach while running like infrastructure. Anything that depended on a human remembering to swap copy per account was going to break by week three, so that was off the table from the start.
That constraint shaped every decision. If something couldn't be templated, triggered, or tracked, it didn't go in.
What I built for personalization at scale
I built it as an account-based system with three layers that fed each other: who we target, what they see, and how it moves through the funnel.
- Segmentation that meant something. I tiered accounts by fit and intent, not just firmographics. High-fit, high-intent accounts got the full treatment, while lower tiers got lighter, automated nurture. The tiering decided budget, channel, and message depth.
- Dynamic content tied to the account. Landing pages and creative adapted to the visiting company and its industry, so a prospect saw their own context reflected back instead of a generic pitch. The aim was a jobs-to-be-done fit: matching the message to the job the buyer was actually hiring a product to do, rather than to their job title. Same template, account-specific substance.
- A connected stack, not a pile of tools. Demandbase and 6Sense handled account identification and intent, Salesforce was the source of truth, and Marketo ran orchestration and lead flow. The point was integration: intent signal in, routed audience out, personalized experience served, activity written back to the CRM.
What made it work was refusing to treat personalization as a content problem when it was really a systems problem. Leaning on the intent-data approach that modern ABM is built around, I let in-market signals drive prioritization, so the plumbing decided who got what, when, and why. The content was the easy part once that was settled.
What happened
The lift didn't come from clever copy. It came from concentrating effort on accounts that were already in-market and giving them an experience that matched where they actually were. We stopped paying to persuade the wrong people and started showing up correctly for the right ones.
I've carried the same logic into enterprise growth work since, applying ABM, lifecycle, and experimentation against a much larger account base, with newer tools like Mutiny, Braze, and Amplitude in the mix. The stack has changed, but the underlying principle hasn't.
What I'd do differently
If I were starting this today, a few things would change.
- Wire up measurement before the first campaign, not after. I spent too long reconstructing attribution that should have been designed in from day one. Decide how you'll prove it works before you build the thing.
- Start narrower than feels comfortable. A tight, deeply personalized program for your top tier beats a shallow one spread across every account. Earn the right to scale by proving it on the few.
- Treat content as modular from the start. Build components that the system assembles per account, instead of bespoke assets you can't reuse. This is where AI changes the math, because what took a content team weeks can now be generated and routed dynamically.
That last point is where most of my current thinking lives. I've written separately about how I build marketing systems with AI tools and my experimentation framework, both of which grew directly out of these early ABM lessons.
Personalization at scale isn't about saying more things to more people. It's about building a system that decides, account by account, what's worth saying at all. When the system is right, the results tend to follow. There's more on how I approach this kind of work on my Sebastian Behar homepage.
