Scope: what this audit covers and what it does not
The question this audit answers is whether the business knows why customers leave and whether its experience, service and follow-up give them reasons to stay, and which weakness costs the most customers. It covers the retention system as a whole: how retention and churnShare of customers at the start of a period who left during it. are measured, why customers leave, onboarding, kept promises and effort, service recovery and support channels, the cancellation experience, win-back, loyalty and lifecycle messaging, and how customer feedback is gathered and acted on.
It does not re-teach the work. Writing a retention plan, with its diagnosis of why customers leave and its four levers, belongs to the customer retention strategy guide. Running lifecycle email week to week is the lifecycle email programme playbook, and recording friction on a walk through the journey belongs to the guide to walking the customer journey.
Email flows, list health and deliverability are scored in depth by the email programme audit; this scorecard takes the email channel as one part of the system rather than rating it again. A store's repeat purchase levers are scored by the online store audit. Where the business needs a map of the stages a customer passes through before scoring, the customer journey mapping framework gives the structure. Every scorecard sits in the audit library.
Before you score: access and evidence
Ask for read access to the customer database or the reports built on it, the support tool where contacts are tagged, the messaging platform that sends onboarding and lifecycle messages, and any survey tool. Without the customer records you cannot rebuild cohorts, so the measurement rows stay provisional until you have them. Fix an evidence windowThe fixed period of data an audit scores, agreed before scoring, long enough to cover a full cycle for the business and kept the same at every rescore. long enough to show several joining cohorts reaching the point where customers usually leave, and keep the same window for every rescore.
Collect:
- retention and churn definitions and the cohort tables behind them;
- contact and cancellation reasons as tagged by support, with volumes over the evidence window;
- a walk through joining, first use, a complaint and cancelling, as a customer would;
- the live onboarding, lifecycle and win-back messages with their triggers and consent basis;
- the loyalty programme terms and the results it reports;
- survey results and the record of what changed because of them.
Write down, before you look at any results, what the business believes are its main reasons for losing customers. Comparing that belief with the tagged reasons is often the most useful finding in the audit.
Criteria and scoring
Rate each row missing (absent or broken), partial (present but incomplete, inconsistent or out of date) or solid (matches the "what good looks like" column). The scale is your judgement against that column, not a number. Rows marked gate decide the band on their own when they score missing: they are the foundations that tell the business where its customers go, and every other row is read through them.
Measuring retention
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Retention read by cohort (gate) | Retention and churn are defined once and read by joining cohort, not by totals | Missing: one blended figure, or definitions that differ between reports / Partial: cohorts built once but not kept up / Solid: one definition, cohorts read on a rhythm |
| Reasons for leaving tagged (gate) | Support contacts and cancellations are tagged by reason, so causes can be counted and ranked | Missing: no tags, or free text only / Partial: tags exist but agents use them inconsistently or cancellations are untagged / Solid: one short reason list, counted and ranked |
| Survey scores tested | Satisfaction or recommendation scores are checked against actual churn before they steer decisions | Missing: scores reported as early warnings with no check / Partial: checked once, never repeated / Solid: tested against what customers did, and used only where they warn |
The churn and retention metrics guide explains why a CohortA group of customers who share a starting event and period, tracked together over time. view beats a blended rate and how to settle what counts as active. Tagging is built in the tutorial on tagging support contacts by reason, and the survey row is scored against the tutorial on testing whether survey scores predict churn.
Early experience
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Onboarding new customers | New customers are guided to first value with a planned sequence, not left to find it | Missing: a welcome message and nothing else / Partial: a sequence exists but first value is undefined or stalls go unnoticed / Solid: first value defined, path planned, stalls spotted and followed up |
| Promises kept | What marketing promises is checked against what customers actually get | Missing: nobody compares claims with the experience / Partial: stated promises checked, implied ones ignored / Solid: a register of promises, each judged on evidence with an owner |
| Customer effort | Common tasks take as little effort as possible, and repeated effort is found and removed | Missing: customers repeat details and chase updates / Partial: some effort removed after complaints / Solid: effort found in records and fixed in ranked order |
Onboarding is judged against the guide to onboarding customers for services and products, kept promises against the tutorial on checking marketing promises against the experience, and effort against the guide to reducing customer effort.
Service
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Service recovery and complaints | Complaints are welcomed, resolved with a fair remedy and used to fix causes | Missing: complaints handled case by case with no record / Partial: resolved well but causes never traced / Solid: a recovery sequence, staff authority to act and a loop back to the cause |
| Support channels and self-service | Customers can reach help in the channels they use, and self-service answers the common questions | Missing: one hard-to-find channel, no help content / Partial: channels offered but standards unkept or help content stale / Solid: channels matched to customers, a help centre that is used |
The service recovery and complaints guide sets out the recovery sequence and states consumer rights as general direction from the Competition and Markets Authority, not legal advice. Channels and self-service are scored against the support channels and self-service guide.
Leaving and returning
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Cancellation experience | Leaving is as easy as joining, any save offer is honest and the door is left open | Missing: cancel by phone or letter only where joining was online, or a maze of screens / Partial: online but hidden, or offers placed in front of the cancel action / Solid: as easy as joining, offers beside it and declinable at once, a clear confirmation |
| Win-back with a stopping rule | Win-back goes to former customers the business may contact, with a reason to return and a rule for when to stop | Missing: lapsed customers mailed with offers indefinitely, consent unchecked / Partial: a sequence with no end, or no reason to return beyond a discount / Solid: contactable leavers only, a reason tied to why they left, a defined end |
The guide to designing the cancellation experience describes a route as easy as joining and notes that regulators such as the CMA treat easy cancellation of subscriptions as a consumer protection matter. It also states, from the ICO's direct marketing guidance, that marketing to leavers needs consent or an equivalent basis and an opt-out. Treat both as general direction, never legal advice. The win-backMessages aimed at customers who have stopped buying or engaging, sent once they pass their usual gap between purchases. row follows the section on win-back and when to stop in the customer retention strategy guide.
Loyalty and lifecycle
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Loyalty programme design | Any loyalty programme rewards the behaviour the business wants, with clear terms and a cost it understands | Missing: rewards on purchases that would happen anyway, terms unclear / Partial: sensible rewards but liability and incremental margin unknown / Solid: a clear job, fair and transparent terms, cost understood (or no programme, deliberately) |
| Lifecycle messaging timed to the customer | Messages follow the customer's stage and behaviour rather than a fixed calendar | Missing: one newsletter to everyone / Partial: some behaviour triggers, the rest calendar-led / Solid: messages tied to states and triggers, with restraint for engaged customers |
A business with no Loyalty ProgrammeA structured promise that rewards customers for continuing to buy, through points, tiers, paid membership or member perks. can score solid on the first row if the absence is a choice; the loyalty programmes guide explains when a programme is the wrong answer and states that terms should be fair and transparent. Lifecycle timing is scored against the lifecycle marketing guide, and the RFM segmentation framework is a common way to define the stages messages respond to.
Listening
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Voice of the customer | Surveys are short, asked at sensible moments and lead to recorded actions | Missing: no surveys, or scores collected and filed / Partial: surveys run but open text unread or the loop never closed / Solid: timed, sampled, read and acted on, with changes recorded |
| Journey walked regularly | Someone walks the journey as a customer on a rhythm and records the friction found | Missing: nobody has walked it / Partial: walked once for a project / Solid: walked on a rhythm, findings logged with owners |
The voice of the customer guide covers timing, open text and closing the loop, and the guide to walking the customer journey sets out the walk itself.
How to score and prioritise
Score the two gates first. If retention is not read by cohort, a growing base can hide a worsening problem and no fix can be shown to work, so the early experience and loyalty rows are provisional. If reasons for leaving are not tagged, the business cannot tell which fix would keep the most customers, and every later ranking is a guess. Mark the affected rows so.
Then rank the remaining weaknesses in this order:
- The reasons that lose the most customers, usually in onboarding, broken promises or service recovery, before any new offer. The tagged reasons and the cohort curves tell you where most customers go.
- The exit: a cancellation that is hard or dishonest risks complaints and the rules, and closes the door on win-back.
- Loyalty, lifecycle messaging and survey refinements next.
- Programme polish last, unless it hides a problem the other rows need to see.
Within each group, rank by the number of customers the weakness touches. A fix in the first weeks, where every new customer passes, usually outranks the same effort spent on a small group of long-standing customers.
Reading the result: what to do with each band
Red means a gate scored missing. Pause new retention offers and loyalty plans, define retention once and rebuild the cohort view, and start tagging contacts and cancellations by reason. Rescore when enough tagged contacts have built up to rank the causes.
Amber means the gates hold at least partially but several other rows are partial or missing. This is the usual outcome. Work a short ordered list, starting with the largest tagged reason for leaving, so the effect of each change shows in the cohorts that follow it.
Green means the gates are solid and most rows are too. Move effort into testing: which onboarding change, save offer or loyalty reward actually changes behaviour. Rescore each quarter to catch drift, and read the LTVTotal value a customer brings over time. trend beside the scorecard.
From findings to a prioritised fix list
Write each partial or missing row as a finding in plain words: what you saw, the evidence, and why it matters. Then turn the findings into one fix list with a line per fix: the finding, the fix, the owner, the guide that explains it and the rescore date. Order the list by the priorities above, not by how easy each fix is.
Retention fixes cross teams, so name the owner carefully. Cohort reporting usually sits with whoever owns the customer data, reason tagging and service recovery with the support lead, onboarding and promises with product or operations as much as marketing, cancellation with whoever owns billing and the account area, and win-back, lifecycle and loyalty with the marketing team. Present the band, the gate findings and the first few fixes to the person who owns the customer relationship, and keep the full scorecard as the working copy for the rescore. The academy exercise on auditing retention and testing win-back offers for a baking supplies shop practises scoring a store's retention system gates first and ranking the fixes, then designing the test that decides between a win-back offer, a message alone and no contact.
Illustrative example
A coffee subscription audits why its base is shrinking
Dabrinzo Coffee, an invented coffee subscription, runs the audit after a year in which sign-ups held steady at around 300 a month while the active base fell from 2,400 to 2,100. Retention read by cohort scores missing: churn is reported as one monthly figure that blends new and long-standing subscribers. Reasons for leaving score missing: cancellation needs a phone call, and the agents taking those calls record nothing about why. Onboarding scores partial: new subscribers get a welcome email and their first bag, then nothing until the next charge. The cancellation experience scores missing, because joining takes a few screens online and leaving does not. Win-back scores partial, with lapsed subscribers mailed a discount every month with no end. A loyalty scheme is planned but not yet built. The band is red on both gates. The fix list puts cohort reporting first, a short reason list for every contact and cancellation second, and an online cancellation with one optional question third, and holds the loyalty scheme. Two months on, the tagged reasons show most leavers going after their second bag and citing too much coffee piling up, so a pause and a frequency change move to the top of the next list.
Common mistakes
- Reading one churn figure. A blended rate can hold steady while recent cohorts leave faster. Score the gate on cohorts, not on the headline.
- Trusting exit reasons at face value. "Too expensive" often means "not worth it to me", which is a value problem. Score the tagging row on whether reasons are counted and read with care, not on whether a form exists.
- Launching a loyalty scheme to fix churn. A programme rewards behaviour; it cannot repair onboarding or service. Score the experience rows before the loyalty row.
- Making leaving hard to protect the numbers. A hidden cancellation produces complaints and closes the door on return. Score it against the route to join.
- Win-back with no end. A sequence that never stops annoys people who have decided and reaches people the business may not contact. Score the stopping rule and the consent basis, not the offer.
- Treating retention as a marketing project. Most causes sit in product, operations and support. Name owners outside marketing on the fix list, or the findings will stay findings.
