Churn that is an onboarding problem wearing a costume.
The pattern
Cancellations cluster somewhere between month four and month nine and get treated as a product problem, or a pricing problem, or a competitor problem. The accounts that leave almost always failed one specific setup action in their first two weeks. Nobody connects the two, because the cause and the cancellation are months apart.
Why it survives a competent team
Because churn gets measured where it is recorded rather than where it is caused. The onboarding dashboard reports completion rates, which look fine, instead of survival split by completion, which does not.
What we check first
Survival curves cut by whether an account completed each first-run action. One action usually separates the curves further than everything else combined, and it is rarely the one the team expects.
What usually changes
Acquisition stops being the thing everyone fixes first. Onboarding gets the attention it was owed, and payback on customer acquisition cost improves without a single change to the ad account.
This is a pattern, not a client result. See all the field teardowns.