A shared map of who is drifting — and what to do next

Use this atlas with your product and growth leads, or walk through it with us during a risk review. It is the custom service model that ties instrumentation to action.

1. Draw the risk bands

   Separate healthy, watch, at-risk, and critical. Define each band with behavioural rules your stack can compute weekly — not gut feel after a board meeting.

2. Name leading indicators

   For each band, list the signals that tend to move first: session depth, billing health, support friction, feature abandonment. Cap the list at three per band.

3. Write the event contract

   Every indicator needs a dictionary entry: trigger, properties, uniqueness, owner. Prediction inputs without contracts rot quietly after the next release.

4. Match responses to bands

   Product nudge, human outreach, light incentive, or no action. Assign an owner and a forty-eight-hour play for each. Measure save-rate and margin, not vanity retention alone.

5. Close the loop

   Weekly atlas review: which bands grew, which responses worked, which labels need a rewrite. Retire noise. Keep the map shorter than the dashboard it replaces.

Where to go next

When you are ready to instrument or audit readiness, start with Churn Risk Instrumentation or a Prediction Readiness Audit. Prefer a structured cohort read first? Book an Early-Warning Cohort Review.