Why the way customers leave shapes whether they return
Most businesses design joining with great care and leaving with none. Sign-up is tested, shortened and measured; cancellation is a phone number, an email address or a page buried in account settings. The reasoning is usually unspoken: a customer who struggles to leave may stay a little longer. The cost of that reasoning shows up elsewhere, in complaints, chargebacksA sale reversed by the customer's card issuer, which can arrive after other checks have closed and should not stay in payment or attribution figures., reviews that warn others off and customers who would have returned but will not.
The customer retention strategy guide shows why leaving deserves design. Its diagnosis of why customers go lists needs that changed, such as a move, a finished project or an outgrown product, as a reason that often has no fix and should be planned for return. Those customers have no complaint. Whether they come back depends largely on how the business treated them on the way out. Its section on win-back adds that returning works best when it is easy, with saved details and no repeated setup, and that a win-back sequence should be short and end cleanly.
The honest persuasion guide states the principle for any site in its section on leaving, cancelling and saying no: cancelling should be as easy to find as the other options, because hiding it produces complaints rather than retention, and a no should be as easy as a yes. It takes that rule from the guide to communicating a price change, which states it for subscriptions when prices change.
There is also a regulatory direction, which the site's guides state at a general level. Regulators such as the Competition and Markets Authority, the UK's central consumer protection enforcer, treat easy cancellation of subscriptions as a consumer protection matter. Their guidance is useful general direction and not legal advice. Contract terms, notice periods, refund rights and anything else that depends on the law in a given country or sector are questions for qualified advisers, and this guide states no rule about them.
This guide covers any business with an ongoing relationship: memberships, service plans, software accounts, contracts and clubs. For an online store's product subscriptions, the subscribe-and-save programmes guide covers clear terms, consent to charges and cancellation for consumables in detail. The customer experience foundations hub covers the journey, onboarding, retention and loyalty that come before a customer decides whether to stay or go.
A route to cancel as easy as the route to join
The simplest test of a cancellation route is to compare it with the route to join, step by step. If a customer can join online in a few screens, they should be able to leave online in a few screens, without a call, a letter or an email to a mailbox that is checked weekly. The table compares the common designs.
| Route | What the customer meets | Effect on the customer | Verdict |
|---|---|---|---|
| Online, from the account, as visible as other settings | A clearly labelled option, a short confirmation, an immediate receipt | Leaves without friction; more likely to return | The default for anything joined online |
| Online, with one optional offer screen | The same route with one declinable screen showing a pause or smaller plan | Fair if declining completes the cancellation in one action | Acceptable when tested, as the save offer test describes |
| Phone only, during office hours | A wait, a script and often a retention pitch | Frustration; complaints; sometimes a chargeback instead | Avoid where joining was online |
| Email or written request | Uncertainty about whether it worked and when billing stops | Repeat contacts asking for confirmation | Replace with an online route and a confirmation |
| Hidden or multi-step maze | Repeated screens, guilt wording, unclear final button | Anger, public complaints, regulatory attention | Never |
A few design rules follow from the comparison.
- Name it plainly. Use "Cancel membership" or "Close account", not "Manage plan" or "Change options", so the customer knows they are in the right place.
- Put it where customers look. In the account settings, in the help centre and in the answer to the question "how do I cancel", each linking to the same route.
- Say what will happen before the final button. The date the service ends, whether any final charge applies and whether anything is refunded, in plain words, so the customer confirms knowing the outcome.
- Finish with a clear end state. A screen and a message that say the cancellation is done. A customer who is not sure it worked will contact support, which is an avoidable contact for both sides.
- Keep a person available. Some customers prefer to talk, and some situations need a person, such as a bereavement or a customer in difficult circumstances. Offer a human route as well as the online one, never instead of it.
Where some customers must leave through a person, for example on a business contract with an account manager, the same standard applies: the request is acknowledged at once, the outcome confirmed in writing, and no one has to ask twice.
Pause, downgrade and save offers without pressure
Many customers who start to cancel do not want to leave so much as stop paying for something they are not using right now. A pause, a cheaper or smaller plan, or a change of frequency can keep them, and it often suits them better than cancelling. The subscribe-and-save guide makes the point for product subscriptions: skip, date change, swap and pause controls keep people who only wanted a change, and they sit alongside an easy cancel button, not in front of it. Account settings that hold these options also hold the move up, and the guide to upgrade paths for existing customers keeps moving down as easy to find there as moving up.
The same holds for memberships and service plans. Offer the alternatives where customers manage their account, long before they reach the cancel step, so fewer arrive there at all. At the cancel step, the options can appear once, beside the cancel route:
- Beside, never in front. The cancel action is visible on the same screen as any alternative, not behind it.
- Declinable in one action. Declining completes the cancellation, with no further screens and no wording that shames the customer.
- Described plainly. A pause states when it ends and what happens then; a smaller plan states its price and what it leaves out; a discount states its end date and the price after it.
- Offered once. A customer who declines is not offered something else on the next screen.
Whether a save offerAn alternative such as a pause, smaller plan or discount shown once to a cancelling customer, beside the cancel route and declinable in one action. is worth showing, and which one, is a decision for a controlled test, not for this guide. The cancellation save offer test randomises customers who start to cancel between direct cancellation and one declinable offer step, reads net retained revenue at a horizon past the end of the offer, and holds complaints and the steps and time to cancel as guardrails. That framing matters: ease of cancelling is a Guardrail MetricA metric an experiment must not worsen beyond an agreed bound, watched beside the primary metric that decides the test., something that must not get worse, rather than a variable to trade for revenue. A save counted at the moment it is accepted can simply delay a cancellation, which the test's horizon is designed to catch.
Confirmation, final bills, refunds and data
The confirmation is the last service message most leavers receive, and it should answer every question they would otherwise ask support. Send it straight away, in the channel the customer used and by email, and include:
- What has ended and when. The last date of service, the last class, delivery or visit, and whether the customer can still use anything until then.
- The final charge. Whether there is one, how much and when, or a statement that nothing further will be charged. If the terms involve a notice period, say what it means for this customer in plain words; the terms themselves are for qualified advice to set.
- Refunds and balances. Whether anything is refunded, how and when, and what happens to credit, prepaid sessions, vouchers or loyalty points. Where a loyalty programme is involved, the loyalty programme guide covers points as a liability, which is why the rule for what happens to them at cancellation should be written into the programme terms in advance rather than decided case by case.
- The customer's data and content. What the customer can download before the account closes, such as files, history or records, and what happens to their account and personal data afterwards. How long each kind of data is kept is a decision for the business's retention schedule, which the guide to deciding how long marketing tools keep personal data sets by purpose, with suppression kept on purpose so an opt-out is honoured if details reappear.
- How to come back. A single line on how to restart, and whether saved preferences will still be there.
Keep the confirmation free of promotion. It is a service message, and the guide to promotional and transactional email explains why messages customers need about their account are kept separate from marketing. A confirmation that doubles as an offer reads as one more attempt to keep someone who has already decided.
One exit question and what to do with the answers
Customers who leave know something the business needs. Ask, but ask once, briefly and optionally. One question, "What is the main reason you're leaving?", with a short list of reasons and a free-text box, placed after the cancellation is confirmed or as an optional step that can be skipped, is enough. A long survey at the exit becomes an obstacle, and a required one becomes part of the maze.
Write the answer options from the reasons the business already hears, using the same wording as the support team's reason list, so that exit reasons and contact reasonsThe main reason a customer contacted support, tagged by the agent from a fixed, defined list so volume and other service metrics can be read by cause. can be read together. The tutorial on tagging support contacts by reason builds that list. Include "my needs changed" and "taking a break" as options, because those leavers are the ones most likely to return and the ones a win-back should treat differently.
Read the answers with care:
- By cohort, not only in total. The churn and retention metrics guide explains why cohort views beat blended rates; exit reasons from customers who left in their first months usually point to onboarding, while reasons from long-standing customers point to value or changed needs.
- Through the jobs-to-be-done lens. The jobs-to-be-done framework asks what progress the customer was trying to make. An exit reason like "too expensive" often means the job the customer hired the service for is no longer worth that price to them, which is a different problem from the price itself.
- As a starting point, not an answer. People often choose the polite option. The jobs-to-be-done framework for retention adds switch interviews with customers who left, which uncover what they did instead and why; those interviews belong to that framework page.
Send the largest reasons to the teams who can address them, in the same review where contact reasons and complaints are read, and record what was changed.
What marketing may send after someone leaves
Leaving changes what the business may send. The site's email guides state the principle from the ICO's direct marketing guidance: marketing messages need consent or an equivalent basis and an opt-out, and regulators separate service messages a customer needs from marketing a business wants to send.
In practice, the rules for leavers are short.
- Service messages continue only as long as they are needed. The confirmation, the final bill, a refund notice and anything else about the closed account. They stay free of promotion.
- Marketing goes only where the customer agreed to it. A customer who signed up to a newsletter separately may keep receiving it; a customer whose only relationship was the membership should not suddenly start receiving offers. Record what each customer agreed to, as the guide to list building and consent asks.
- Opt-outs go on the suppression list. Anyone who opted out of marketing, at any point, goes on the Suppression ListA list of addresses or contacts excluded from every send, such as unsubscribes, hard bounces and spam complaints. so they are not reached again if their details arrive from an import or a sync.
- Win-back is short and ends. Where permissions allow a Win-Back CampaignMessages aimed at customers who have stopped buying or engaging, sent once they pass their usual gap between purchases., the retention strategy guide's rules apply: lead with what has changed, make returning easy, use an incentive only as a final nudge, and stop after a defined sequence.
- Tell customers before you treat them differently. If leavers will receive a small number of messages after cancelling, say so in the cancellation flow and give a one-step way to say no.
Keep track of how these decisions play out. Cancellation complaints, unsubscribes from leavers and the share of returners who left for each reason are the measures that show whether the leaving experience is working. The Churn RateShare of customers at the start of a period who left during it. itself is read in the churn guide; this experience is judged by how people leave and whether they come back. When a win-back is already running and few leavers return, the diagnosis for win-back campaigns that are not bringing customers back counts a hard or hostile exit among the causes it checks.
A cookery club redesigns its cancellation (illustrative)
Skorvuth Cookery Club replaces its cancel-by-phone rule
Skorvuth Cookery Club, an invented monthly cookery class membership with about 2,300 members, lets people join online in three screens but cancel only by phone during office hours. Cancellation calls make up 18% of all support contacts, and 11 complaints a month mention how hard it is to leave.
The club adds a "Cancel membership" option to the account page, beside a "Pause for up to three months" option. The cancel route has one screen that shows the pause once, with "No thanks, cancel" as prominent as "Pause instead", and declining completes the cancellation. The confirmation email states the last class, the final charge of £34 on the next billing date, that unused class credits remain bookable until the last class, and how to restart with saved dietary preferences.
One optional question follows: "What is the main reason you're leaving?" In the first quarter, 38% of leavers choose "busy period" or "taking a break", 24% "cost" and 19% "moved away". Leavers who agreed only to membership emails are added to the suppression list for offers; those subscribed to the recipe newsletter keep it.
Over the next two quarters, cancellation complaints fall to 2 a month and cancellation calls disappear from the support queue. Of members who chose a pause, 61% resume when it ends, and 9% of members who cancelled for "busy period" or "taking a break" rejoin within six months.
Common mistakes
- Making leaving harder than joining. A phone-only or hidden route trades a short delay in cancellations for complaints, chargebacks and customers who never return.
- Putting offers in front of the cancel route. Pauses and smaller plans help when they sit beside cancellation and can be declined in one action; in front of it they are an obstacle.
- Judging save offers at the moment they are accepted. Many only delay a cancellation. Test them properly, with ease of cancelling as a guardrail.
- A vague confirmation. Leaving out the last service date, the final charge or what happens to data and credit creates repeat contacts and distrust.
- Selling in the confirmation. The last service message is not a marketing opportunity.
- A long or required exit survey. One optional question gets better answers and keeps the exit clean.
- Reading exit reasons in total only. Cohorts and the jobs-to-be-done lens show what the reasons mean.
- Marketing to leavers without permission. Send only what customers agreed to, and keep opt-outs on the suppression list.
- A win-back that never stops. A short sequence with a defined end respects the decision and keeps the door open.
- Treating legal questions as design choices. Notice periods, refund rights and contract terms need qualified advice in each market.
