Subscription Retention Audit: 30-Point Worksheet

Most subscription businesses lose subscribers in places nobody is watching: a retry schedule nobody has reviewed in a year, a dunning email that still names an old plan, a cancel flow that asks nothing and saves no one, a winback that was never built. This worksheet turns "we should probably look at retention" into a scored audit you can finish in one sitting. Thirty questions, five sections, each scored 0, 1, or 2 — no, partially, yes. Answer honestly: a flattering score on a broken system helps no one, and the point of the exercise is the fix list it produces, not the number.

Section 1: Payment recovery

1. Failed payments are retried automatically rather than expiring on the first decline.
2. Retry timing is based on the decline reason, not a fixed interval for every failure.
3. Hard declines (stolen card, closed account) are never retried.
4. Multiple retry attempts are spread across days, not fired in a single burst.
5. A backlog of previously failed payments was worked when your current system started, not written off.
6. Card updater services refresh replaced card numbers before a charge is attempted.

Fix first: if questions 1–3 scored low, nothing downstream matters much — recovery starts with decline-aware retries. The full method is in the failed payment recovery guide.

Section 2: Dunning communications

7. A failed payment triggers an email to the subscriber the same day it fails.
8. The sequence has more than one touch — a single email is not the whole program.
9. At least one touch reaches the subscriber by a second channel, such as SMS.
10. Every email links to a card-update page that opens without a password hunt.
11. The card-update page carries your name and brand, states the amount, and confirms the fix.
12. The final notice, before cancellation, says plainly that the subscription is about to end.

Fix first: the link and the page. Emails that send motivated subscribers to a login wall or an unbranded form waste the touches you already have; fix the landing before adding sends.

Section 3: The cancel moment

13. Clicking cancel opens a flow, not an instant termination.
14. The flow asks why the subscriber is leaving, in their words or from a short list.
15. At least one save offer is presented before the cancellation completes.
16. Offers match the stated reason — a price objection and a "too busy right now" get different answers.
17. The offers include more than a discount: a pause and a service or help option, for example.
18. Completing the cancellation is still easy and honest — no dark patterns, no forced calls.

Fix first: if the flow saves no one because it offers nothing, start with reason-matched offers; the mechanics live on the cancel deflection page. Swipe Again's own deflection, included in its flat price, works with Discount, Pause, and Service offers for reference on the shape.

Section 4: Measurement

19. You can state last month's payment failure rate without opening a report builder.
20. Recovery rate is tracked separately from overall churn.
21. Recovered revenue is a number someone reviews monthly, in dollars.
22. Involuntary churn is reported separately from voluntary cancellations.
23. You know what your recovery program costs per dollar it recovers.
24. Someone owns these numbers — a named person checks them on a schedule.

Fix first: definitions before dashboards. Agree on what counts as recovered and when the clock starts, write it down, and pull the same numbers the same way each month — direction against your own baseline beats borrowed benchmarks.

Section 5: Winback

25. Subscribers lost to failed payments are identified as a distinct group, not merged with voluntary churn.
26. That group receives at least one reactivation attempt after the loss.
27. The winback offer acknowledges what happened — a payment problem — rather than pretending they chose to leave.
28. Returning subscribers can reactivate without re-entering everything from scratch.
29. Winback results are counted separately, so you know whether the effort pays.
30. There is a defined window after which winback attempts stop and the record rests.

Fix first: question 25. Until payment-lost subscribers are a separate list, every winback is a generic blast; segmenting the list is the one move that makes the rest of this section possible.

Scoring bands

Total score	Band	What it means
50–60	Systematic	Retention runs as a system: recovery, communication, the cancel moment, measurement, and winback all exist and connect. Your work is tuning — timing, copy, offer mix — and watching the trend.
38–49	Built, with gaps	The machinery exists but at least one section is hollow. Find your lowest section and close it before optimizing the strong ones; the gap is where the losses concentrate.
24–37	Partial	Some pieces run — usually retries and a first email — while the cancel moment, measurement, or winback is missing entirely. You are recovering the easy share and losing the rest silently.
12–23	Reactive	Retention happens by accident: a native retry here, a support save there, no owner and no numbers. Almost any structured fix pays for itself; start at Section 1 and work down.
0–11	Exposed	Failed payments and cancellations run their course untouched. The honest read: a meaningful share of your churn is not a product verdict at all — it is process absence, and it is the most fixable kind.

What to do with your score

Work section by section, lowest first, and resist polishing a section that already scores 10 or 11 while another sits at 3. Re-run the audit after each fix ships — the score is a tracking tool, not a ceremony — and keep the dated worksheets so the trend is visible. If you are building the program from a low base, sequence the work with the dunning launch checklist, which orders the same fixes into a launch plan. And one number to keep in view while you work: 20–40% of subscriber churn is involuntary (Recurly research). However your sections scored, that share of your losses is the part this worksheet exists to shrink.

The bottom line

Retention rarely fails in one dramatic place; it leaks across five ordinary ones. Thirty questions, scored without mercy, show you which leak is yours. Fix the lowest section first, re-audit, and repeat — that loop, run quarterly, is most of what a retention program actually is.

Frequently asked questions

How long does this audit take?

About thirty minutes if the answers are within reach, and that qualifier is the point: questions you cannot answer without a research project are themselves findings, usually in the measurement section. Score those honestly, note what it took to find the answer, and make the number easier to reach next quarter.

Who should complete the worksheet?

Whoever owns subscription revenue, with input from whoever runs billing operations. One person should own the final scores and the fix list. Audits completed by committee tend to average away the disagreements, and the disagreements are usually where the truth is.

How often should we re-run the audit?

Quarterly, and again after any major change: a new billing platform, a pricing change, a new cancel flow, or a new recovery tool. Keep the dated worksheets. A single score is a snapshot; three scores are a trend, and the trend is what tells you whether fixes are holding.

What if several sections score equally low?

Fix them in worksheet order: payment recovery, then dunning communications, then the cancel moment, then measurement, then winback. That order follows the subscriber's path — a payment has to fail and be worked before communications matter, and winback only makes sense once the earlier stages stop producing avoidable losses.

Does a high score mean our churn will drop by a specific amount?

No, and any worksheet that promises a specific result is inventing it. The score measures whether the retention system exists and connects, not what it will return; returns depend on your decline mix, price point, and audience. What a rising score does guarantee is that fewer losses happen for process reasons — the category of churn most fully in your control.