Sinch

The Problem

Marketing teams didn’t trust their own data. It was scattered across systems that didn’t talk to each other, which meant every targeting decision and every ROI number came with an asterisk nobody said out loud. That’s not really a data integration problem. That’s a “can I actually believe this number” problem — and those are much harder to fix.

My Role & Scope

I owned discovery and definition for a unified customer data platform, working across engineering, data, and business stakeholders inside a genuinely messy enterprise environment: legacy systems, a long list of stakeholders, and zero tolerance for getting it wrong.

Key Product Decisions

I went after the signals that actually moved targeting and ROI, not full data completeness — chasing every possible data source first is how these projects die before they ship anything real.

I chose incremental rollout over one big migration. Each phase had to earn trust on its own, because in an enterprise, trust is the actual currency — nobody adopts a platform they don’t believe in yet.

And I kept the models explainable even when a more sophisticated approach was available, because a recommendation nobody on the business side understands is a recommendation nobody acts on.

Execution Highlights

I defined the data models and ingestion approach with the platform and data teams, and built toward the specific downstream use cases — targeting, attribution, measurement — that marketing teams actually needed, not the theoretically complete version of the platform.

What Actually Happened

ROI confidence improved. Targeting got more accurate. The foundation held up well enough to support what came after it.

What I took from this one: enterprise data products don’t fail on the technology. They fail because nobody designed for trust and adoption on purpose — they assumed it would follow from the data being “correct.” It doesn’t. You have to build it in.