04 · Renew or leave
A churn date records the moment a customer told you. The decision was made earlier, in ordinary use, while the account still looked healthy enough not to ask.
Definition
A Churn Decision Study is a customer research study that reconstructs where a customer's decision to leave began. It interviews cancelled accounts, accounts reducing use, and healthy renewals to recover the pains that accumulated in ordinary use, the triggering event, and what the customer hired instead. It explains the churn rate, which records the date notice arrived rather than when the decision formed.
By the time notice arrives, the customer has already reconstructed their own situation, weighed alternatives and made a choice. Exit surveys interview them after the argument is settled, which is why the answer is so often price.
Your churn rate observes stage six. The study reconstructs stages one through five.
Economics
This decision is the one that works against revenue you already have, so it is modeled against the installed base rather than a new cohort.
Current ARR book × annual gross revenue churn rate
Gross rather than net, so expansion does not mask what is leaving.
Some of this book was never a good fit, and a study that establishes which part is still a useful result. It is more often cheaper to keep revenue than to replace it, but that is a claim to test against your own numbers rather than assume.
Break-even framing in percentage points of movement. Method does not change the size of that movement. It changes how much of the explanation you can trust, and therefore which decisions you can defend making from it. It is not a projected return, and pricing shown is a floor and a typical range, scoped per engagement. See how the three methods differ.
The design
Interviewing cancelled accounts alone tells you about the end state. The customers still paying while quietly reducing use are the ones who can still be reached, and they are living the middle of the same story.
The complete arc, including what they hired instead and whether it solved what they thought it would.
Still paying, using less. The decision is forming now and is still observable in progress rather than in hindsight.
The control. What kept the value registering for them, which is rarely the feature list anyone expects.
What the study resolves
Business implications
Identify the observable behaviors that precede the decision, so health scoring reflects what actually predicts leaving.
Address the pains that accumulate quietly in ordinary use, and the points where delivered value stops being legible.
Intervene while the decision is still open rather than at notice, and stop buying renewals with discounts.
Interview evidence generates hypotheses about why accounts leave. Check them against usage, support and account data before rebuilding a health score on them.
Engagement outline
Review churn definitions, recorded cancellation reasons and the current health score.
Select recently cancelled accounts, accounts reducing use, and healthy renewals for contrast.
Follow accumulated pains, the triggering event, the search, and the moment something else was hired.
Deliver patterns, exceptions, the signals that were observable, and predictions to test in account data.
Synthetic, AI-moderated and 1:1 expert execution options are compared on the methods page.
Fit
Best suited to Seed to Series B B2B SaaS companies with an installed base experiencing:
If customers are leaving before they ever put the product to real work, this is an adoption problem wearing a churn costume. Start with the Adoption Reality Check instead.
Common questions
Because they arrive last. By the time a customer files notice they have already settled the argument internally, and price is the most available and least awkward way to summarize it. The decision formed earlier, during ordinary use, and reconstructing that period is what produces a different answer.
Well before notice, usually while the account still looks healthy. The common sequence is pains accumulating in daily use, delivered value ceasing to register, a triggering event, an active search, then hiring an alternative. Cancellation is the sixth stage, and it is the only one most churn reporting can see.
Yes. Accounts that are quietly reducing use are living the middle of the same story and can still be reached, which cancelled accounts cannot. Adding healthy renewals as a control shows what kept value registering for them, which is rarely the feature list anyone expects.
Sometimes, and it is worth ruling out first. If customers are leaving before they ever put the product to consequential work, nothing was retained to lose and the decision was made much earlier than the renewal date. That pattern points to an Adoption Reality Check rather than a churn study.
Start with one account you lost
Bring your gross churn rate, your recorded cancellation reasons and the account whose departure nobody saw coming.