Revolut
In the early years, Revolut was growing rapidly in registered account numbers — but the analytics team had discovered a worrying pattern: a significant portion of users never used the card after activation.
Cohort analysis revealed a critical threshold: users who made at least 3 transactions in the first month had dramatically higher 12-month retention than those who made 1-2. It was not linear — it was a discontinuity. Three transactions represented the point at which the user had integrated Revolut into their daily behaviour.
This redefined the product team's north star metric: not "new accounts registered" but "activated accounts" (at least 3 transactions in month 1). Every feature, every notification, every onboarding flow was evaluated for its impact on this metric.
Practical consequences: Revolut invested heavily in onboarding that actively pushed users to make the first transaction within 24 hours of activation — often offering cashback or a bonus for the first payment. And it reduced friction in the card-adding process, which was the step many users did not complete.
The most important KPI is not always the most visible. Finding the behavioural threshold that predicts long-term retention is more valuable than optimising the most obvious growth metric.