By the Time They Give Notice, You've Already Lost

Save rates collapse at the notice stage

Once notice is served, expect to save 10–15% of accounts. Intervene ninety days or more before renewal on a leading indicator and save rates run 40–60%.

On €10M of recurring revenue with 20% churn, €2M is at risk each year. Moving the save rate from 12% to 40% retains around €560k annually, and almost all of it lands in EBITDA.

The signals that actually predict

Usage or login frequency declining over a sixty-day window. Support tickets going to zero — silence, not complaints. The departure of your internal champion, which carries two to three times the churn risk within two quarters. Payment drifting from fifteen days to forty-five. A skipped quarterly review. A narrowing set of services or features consumed.

Build these into a health score from zero to a hundred, then backtest it against twenty-four months of real cancellations. If it does not retrospectively predict who left, it is astrology with a dashboard on top.

Why exit interviews mislead you

Owners believe churn is a price problem because that is what the exit interview says. Exit interviews are polite. "Too expensive" is the socially acceptable version of "you never delivered anything I could point at".

The real cause usually sits four to six months upstream, in an onboarding that never produced a visible win. And the intervention has to be proof of value — the report showing what they received — not a discount. Discounting at the churn moment teaches your entire base to threaten.