From Zero Personalization to $300M in Monthly Volume

Seventy million users, and every single one of them saw the same thing. Not “bad personalization” — literally no personalization. No recommendation models. No segmentation beyond basic demographics. No A/B testing framework. The data science team had talent but no product strategy to aim at. Engineering had the platform but no requirements. Everyone agreed the product should feel more personal. Nobody had defined what that meant. The data scientist who’d been hired to build recommendation models was running exploratory analyses that went nowhere — not because she lacked skill, but because nobody could tell her what problem to solve. The PM responsible for engagement couldn’t explain why metrics were flat despite a product 70 million people used daily.

The product was succeeding on network effects and brand alone. Which meant massive value was sitting on the table, waiting.

Defining Personalization for a Social Payments App

The temptation was to import a personalization playbook from a content platform — what works at Netflix or Spotify. I didn’t do that. A social payments app isn’t a content platform. What “personal” means is fundamentally different. So before building anything, I had to admit I didn’t have the answer and go find it.

I built the wireframes myself and partnered with data science to design propensity models — not just “who is this user” but “what is this user likely to do next, and what would actually be useful to them right now?” That distinction matters. Generic personalization blasts recommendations at people. Useful personalization anticipates need.

The Insight Nobody Expected: Communication Fatigue

Here’s what I discovered: after years of sending everyone everything — every promotion, every notification, every banner — customers had developed communication fatigue. Banner blindness. Notification numbness. They’d stopped paying attention entirely. The problem wasn’t that we weren’t personalizing. The problem was that we’d trained 70 million people to ignore us.

That changed the strategy completely. This wasn’t about adding recommendations on top of what existed. It was about earning attention back. Once we understood segmentation better and what each customer’s propensity was for different products and services, we could be relevant instead of noisy.

Small Wins, Not Grand Strategy

I pushed for something the team wasn’t used to: a series of small, fast experiments instead of one big bet — the same discovery sprint approach I use on every engagement. Get the first models live within two to three months. Measure. Adjust. Ship the next one. Every feature had to scale with growth while maintaining latency SLAs — personalization that slows down the app is worse than no personalization.

We launched personalized discovery and feed enhancements. Each one was a small win on its own. But they compounded. I trained the data science team on the product strategy framework so they could run the next round of experiments without me — the same developing-the-team-not-just-solving-the-problem approach I bring to every engagement.

Results: 319% Conversion Lift and $300M Monthly Volume

First-time conversions lifted 319%. The incremental payment volume hit $300 million per month. Click-through and engagement improved platform-wide. What started as “we should do personalization someday” became a core product capability.

I remember the meeting where the first real numbers came in. The room went quiet. Someone on the data science team — the same person who’d spent a year running analyses with no direction — looked at the 319% conversion lift and said, “So this is what happens when we actually know what to aim at.” That’s the moment personalization stopped being a roadmap item and became real.