Cohort Experimentation

At Akseleran, the activation loop was broken. Users were signing up but not completing first investments. The conventional fix was a blanket discount. I designed a cohort-based cashback experiment instead — targeted at users showing specific behavioral signals, not the full funnel.

The retention impact was 50 percentage points. The GTV impact was $2.7M incremental. More importantly, the unit economics held — because the targeting meant we weren't subsidizing users who would have converted anyway.

The work also produced a framework for cohort experiment design in fintech: how to isolate behavioral signals, how to structure incentive tiers, and how to measure incrementality rather than correlation.