Churn Analysis Template

Begin Before You Explain the Loss

An account cancels and selects “too expensive.” The tempting response is a discount, a win-back message, or a cheaper plan. Yet the account stopped completing the core job three weeks earlier, after two failed imports and a founder-assisted workaround. Price may be the final sentence without being the cause of departure.

Use this review to reconstruct what happened before churn, compare that path with other lost and retained accounts, and choose one intervention. It is not a win-back script. A candid loss is more useful than a rescued account whose underlying problem remains hidden.

Fix the Question and the Denominator

Run the review after a meaningful cluster of departures, a lost target account, an unexpected revenue contraction, or a change in churn inside one segment. Declare the customer’s clock before calling an absence churn. A monthly-close product cannot lose an account in a week simply because the next close has not arrived.

CHURN REVIEW BOUNDARY
Decision this review must inform:
Target segment and recurring job:
Retention unit: person / account / workspace / site / other
Natural value cycle:
Churn event: cancellation / non-renewal / failed payment / missed cycles / contraction
Observation window:
Accounts eligible to return or renew:
Accounts that reached first value:
Lost accounts and recurring revenue lost:
Comparison group of retained accounts:
Known gaps in events, billing, or customer records:

Keep exits before first value separate from losses after value. The first group usually tells you about qualification, promise, trust, setup, or activation; the second can speak to recurring utility, reliability, workflow fit, price, or delivery burden. Separate an involuntary billing failure from a customer decision until the evidence joins them. Likewise, keep account loss beside revenue loss. One large cancellation should not be allowed to impersonate a broad retention pattern, and several small cancellations should not disappear behind one expanding account.

Rebuild One Account’s Path

Start with evidence close to the account: product events, billing changes, support threads, founder interventions, and the customer’s own explanation. Put them in time order. A cancellation reason interpreted without the preceding behavior is only a label.

LOST-ACCOUNT TRACE
Account and declared segment:
Buyer, operator, and recurring job:
Acquisition source and promise that brought them in:
Started / activated / ended:

First value event and time to value:
Value cycles eligible / completed without help / completed with help / missed:
Last successful value event:
First visible break in the path:
Errors, delays, trust concerns, or support history near the break:
Founder minutes, manual steps, or custom work supplied:

Customer's exact departure language and source:
Alternative chosen or old workflow resumed:
Price, plan, renewal, or contraction change:
Observed facts:
Customer-reported facts:
Founder's inference:
Missing evidence worth pursuing:

If the account will talk, ask about the work rather than defending the product: What were they trying to complete? When did they first expect the product to help and it did not? What did they do next? What are they using now? What would have needed to be true for the product to remain part of the workflow? Record the answers against the trace; do not negotiate during the inquiry.

Locate the Earliest Broken Transition

Read the trace from arrival to departure and stop at the first transition the evidence cannot support:

1. A qualified customer recognized the promise.
2. The account committed and reached first value.
3. It repeated that value when the job returned.
4. The product delivered reliably without hidden founder rescue.
5. The buyer accepted the price and renewal terms for that value.

The earliest break constrains the diagnosis. An account that never reached first value cannot prove that recurring value was weak. A retained account whose renewal failed after a surprise price change says something different from an account that had already abandoned the workflow. A customer who returned only because the founder cleaned every import may prove pain and willingness to collaborate while rejecting the current product economics.

Name a likely cause only after locating the break. Then write a competing explanation. “The importer destroyed trust” competes with “this segment closes books too infrequently to form a recurring habit.” “The plan was too expensive” competes with “the buyer never saw the result the operator received.” The purpose is not perfect attribution; it is to prevent the most convenient story from becoming a roadmap.

Compare Losses With Accounts That Stayed

One loss can expose a defect or trust failure. A product decision usually needs a pattern. Compare lost accounts by segment, recurring job, acquisition source, activation path, value cycles, founder assistance, price or plan, and chosen alternative. Then inspect retained accounts with the same characteristics. Retained customers are the control that a cancellation form cannot provide.

CHURN PATTERN REVIEW
Lost accounts traced:
Comparable retained accounts reviewed:
Shared earliest break among losses:
Difference visible in accounts that stayed:
Losses that never activated:
Losses after repeated value:
Losses preceded by product or trust failure:
Losses preceded by recurring founder rescue:
Losses concentrated in one segment, source, plan, or product version:
Customer explanations that match observed behavior:
Explanations contradicted by observed behavior:
Strongest counterexample to the current diagnosis:
Narrowest conclusion the evidence supports:

Suppose a compliance-evidence product loses four of twelve small vendor accounts at renewal. Two never completed source setup. A third produced monthly evidence twice, then returned to a consultant after an export omitted audit links. The fourth used the product only when the founder classified documents by hand. Meanwhile, six retained accounts all completed setup with the same three source systems and generated traceable exports without assistance.

“Four accounts found the product too expensive” would produce the wrong work. The losses reveal three distinct boundaries: qualify or repair setup for the first two, harden export traceability before asking similar accounts to depend on it, and price or refuse the document-classification service instead of hiding it inside the subscription. A blanket discount tests none of them.

End With One Bounded Response

Choose the response that addresses the strongest pattern while adding the least new surface area. That may be to narrow the segment, repair one failed transition, change onboarding, harden a relied-on path, repackage a service, contact more lost accounts, or stop serving a weak-fit group. Do not combine all plausible causes into one retention project.

CHURN DECISION NOTE
For [segment], the path to churn first breaks at [transition].
The best evidence is [behavior, customer words, and comparison].
[Competing explanation or counterexample] limits this conclusion.
We will [one intervention] for [bounded accounts or cohort].
By [next natural value cycle or renewal], we expect [observable change].
We will reject this diagnosis if [disconfirming result].
We will not yet [larger feature, discount, campaign, or segment expansion].
Owner and review date:

The review is complete when the loss has changed a decision and the next cohort can contradict it. If the trace is mostly blank, the next action is not a feature. Restore enough account, event, support, billing, and founder-work history to learn from the next departure.