Lean Analytics: Use Data to Build a Better Startup Faster

by Alistair Croll

We recommend defining an inactive user as someone who hasn’t logged in within 90 days (or less). At that point, they’ve churned out; in an always-connected world, 90 days is an eternity.

Churn Complications

Worse: if you’re counting churns as “someone who hasn’t come back in 30 days,” then you’re comparing last month’s losses to this month’s gains, which is even more dangerous, because you’re looking at a lagging indicator (last month’s bad news). So by the time you find out something is wrong, it’ll be next month.

Ultimately, the math gets really complex. There are two ways to simplify it. The first is to measure churn by cohort, so you’re comparing new to churned users based on when they first became users. The second way is really, really simple, which is why we like it: measure churn each day. The shorter the time period you measure, the less that changes during that specific period will distort things.