“If you look at it from the framework of LTV of a cohort that's probably the right framework as a starting point… that's a hard thing to optimize for in a single experiment — you have to project based on things that are happening at day eight or day 30.”

Cohort LTV Is the True North — Proxy Metrics Get You There Faster

Burner runs paywall and pricing experiments with cohort LTV as the north star, but since maturing LTV data takes months, the team uses early proxy signals (day-8 conversion, day-30 retention) to pick winners before the full curve is visible. Knowing what you're optimising for upfront is the prerequisite for trustworthy experiments.