Free tool • 30 seconds • Your data never leaves the page
Health Score Back-tester
Would your health score have caught your churn? Paste your score history and get the answer in 30 seconds: your catch rate, your false alarm rate, and an honest read on whether your CSMs should trust the number.
A score is a prediction. This grades the prediction.
Step 1 — Your data
Paste your score history
Three columns: account, health score (0–100, or red/yellow/green), status (churned or active). Use the score as it stood ~90 days before each outcome. Everything runs in your browser — your data never leaves this page.
How to read your result
Catch rate is the only number that matters
Of the accounts that churned, what share did your score flag in time? Everything else — model sophistication, number of inputs, dashboard polish — is decoration if this number is low.
False alarms are why CSMs stop looking
A score that cries wolf trains your team to ignore it. The threshold slider shows the trade-off directly: every score buys catch rate with false alarms. The art is spending that budget where churn actually lives.
The misses are the roadmap
Pull the list of churned accounts your score missed and ask what they had in common — quiet champions, stalled onboarding, sponsor changes. Those patterns are the inputs your next score needs.
Questions
What exactly does a back-test prove?
Whether your score would have flagged the accounts that actually churned — before they churned. A score is a prediction; a back-test grades the prediction against history. If it cannot predict your past churn, it will not predict your future churn.
What data do I need?
Three columns: account, health score (0–100, or red/yellow/green), and status (churned or active). Use each score as it stood roughly 90 days before the outcome. A dozen churned accounts is enough for a meaningful read; more history is better.
Is my data uploaded anywhere?
No. The entire test runs in your browser. Nothing is sent to any server — close the tab and it is gone.
What is a good catch rate?
In a 2026 poll of CS leaders, 82% said their health score caught less than half of churn — so anything above 50% already beats most teams. Above 75% is strong. Below 50% means the score is measuring the wrong things, usually activity instead of value.
My score failed the test. Now what?
That is the normal outcome — and the fixable one. We design and back-test replacement scores on your stack as a fixed-price engagement.
Fixed scope. Fixed price. Built on your stack.