How to reduce churn and handle cancellations

1 The bottom line

Diagnose why people leave, then fix the early experience, save them at the cancel moment, and recover failed payments. Most churn comes from weak value and early exits, and cancelled users are hard to win back.

2 Key takeaways

• Treat churn as a value problem first. Users leave because the product is not worth the money, not because a rival is better.
  • Cost plus low usage make up about 70% of Google Play cancellation reasons Jacob Eiting and David Barnard, RevenueCat50:00.
• Win retention early. Early churn is the most common and the most expensive.
  • 34% of annual churners cancel in the first month Jacob Eiting and David Barnard, RevenueCat59:00.
• Build a save step into the cancel flow. Once people turn off auto-renew, they rarely come back.
  • A seven day trial extension beat a 30% discount and a 3 month pause Brett Bauman and Zack Hargett, Coconote51:35.
• Recover failed payments. Some cancellations are not choices at all.
  • Nearly a third of Google Play cancellations are involuntary billing failures Jacob Eiting and David Barnard, RevenueCat47:27.
• Have retention tools ready before a spike. Press and awards bring users you will lose without them.
  • Pok Pok won an Apple Design Award with no save flow and lost many new users Melissa Cash, Pok Pok16:25.

you've got to ask when people are in trouble but not yet canceled. You might be able to save them.

Jason Cohen, WP Engine

2.1 Size the problem and find the real reason

Put a number on the cost of churn, then get honest answers about why it happens.

• Compute your customer ceiling. Jason Cohen, WP Engine16:26.
  • Divide the customers you add each month by your monthly cancellation rate.
  • Cancellations grow with your base, but marketing output does not, so growth slows near the cap.
  • Adding 100 customers a month at 5% monthly cancellation caps you at 2,000 customers.
• Ask "what made you cancel" in an open-ended email. Jason Cohen, WP Engine19:53.
  • Use free text, not a dropdown.
  • The wording pushes people to describe what in the product or their situation caused it, not a one word answer like budget.
  • Groove's same email went from 10% usable responses to 20% after changing "why" to "what made you".
• Randomize multiple-choice reasons or drop them. Jason Cohen, WP Engine19:06.
  • People pick the first or easiest option, so the data is noisy.
  • At Smart Bear the first item was picked most. After randomizing, all items were picked equally.
• Dig past "too expensive" to the deeper cause. Jason Cohen, WP Engine20:53.
  • Customers saw your pricing page and chose to buy, so price was acceptable at purchase.
  • Look for a missing integration, an unfulfilled promise, or a badly chosen target segment.
  • A customer who needed a Jira integration but used Linear listed price. The real gap was the integration.
  • Even "project ended" can mean the product was not useful enough, or that the segment, like small business, has high variance.

2.2 Fix the first days and the first month

Most cancellations come early, so early value does the most to keep people.

• If you are unsure where to start, work on onboarding. Jason Cohen, WP Engine27:06.
  • Most cancellations happen in the first day, 30 days or 90 days.
  • Early churn is the most unprofitable because acquisition cost is never paid back.
  • Like a YouTube retention curve, moving 50% to 55% staying at the front could move final retention from 20% to 25%.
• Deliver value in the first month of an annual plan. Jacob Eiting and David Barnard, RevenueCat59:00.
  • Annual churners mostly turn off auto-renew soon after the charge, not just before renewal.
  • Month 11 has a bump, rising from 4.7% to 11% at renewal, but that is small next to month one.
• Track day zero cancellations if you run a free trial. Anmol Tiwari, Duolingo24:25.
  • The share of users who cancel right after starting shows how high their intent is.
  • Catch the problem early, because un-cancelling is very hard.
  • The host adds that paid trials can give ad networks a cleaner signal than free trials, which people turn off immediately.
• Pair one big saving with many small growth features. Jimmy Liu, DoorDash37:00.
  • The core benefit must be large. DashPass saves roughly 80% on fees, not 10 or 20%.
  • Small features in signup, billing and discovery each add a point or two and add up.
  • Jimmy says the resulting retention is causal, not just correlation.

2.3 Reach out before they cancel

Customers in trouble can still be saved, so find them while they are reachable.

• Watch for warning signals and contact those customers. Jason Cohen, WP Engine25:55.
  • Signals include never uploading data, too many or too few support calls, and no logins.
  • You can do this with few customers. Form a theory, act on it, and adjust as data grows.
  • When comparing good and bad customers, look for traits that differ between the groups, not traits both share.

2.4 Save them in the cancel flow

Offer more time at the moment of cancelling, and build the flow before you need it.

• Offer seven more days to trial users who try to cancel. Brett Bauman and Zack Hargett, Coconote32:30.
  • Ask whether they just need more time, instead of only showing lost value and asking questions.
  • Trial users are hard to win back once they turn the trial off, so keeping it enabled matters.
  • The extension saved about 27% of people going through the cancel flow, though a smaller share go on to convert.
• Test the offer against alternatives. Brett Bauman and Zack Hargett, Coconote51:35.
  • They tried the extension against 30% off and a 3 month summer pause.
  • The extension won by far, and web cancellation flows kept 25% of people from cancelling.
• Add save tools before a big award or feature spike. Melissa Cash, Pok Pok16:25.
  • Pok Pok had no unsubscribe save flow and one global price when the award hit two weeks after launch.
  • Many award users stayed as early adopters, but many were lost.

2.5 Recover failed payments

Involuntary churn is a fixable slice of cancellations, especially on Android.

• Turn on grace periods and billing retry on Google Play. Jacob Eiting and David Barnard, RevenueCat47:27.
  • Causes include prepaid cards running out of balance and no auto-recovery path like iOS.
  • Tell users about payment problems.
  • This matters most once Android is more than experimental revenue.

2.6 Win users back and ride the seasons

People who left are still your best audience when demand shifts or you launch something new.

• In the off-season, use pause plans, surveys and reactivation moments. Edward Tian, GPTZero32:30.
  • Education demand peaks at back to school and exam season, so work both windows and reduce summer churn in between.
  • Shift users off manual plans.
  • Daily actives swing from 150,000 to a million or more in exam season.
• Take new offerings to churned and registered users first. Mike Maser, Zero46:10.
  • Zero plans to bring its GLP-1 offering to past users, including those who left because they went on a GLP-1.
  • Early users also give feedback on what they like and dislike.
  • Zero has 10 million registered users, an owned audience with 8 years of brand trust that a new entrant cannot match.

2.7 Where they disagree

Sources differ on whether price is a real reason, whether surveys can be trusted, and whether to show premium features.

• Is cost a real churn reason?
  • Jason Cohen, WP Engine20:53 says "too expensive" is never the real reason.
  • Jacob Eiting and David Barnard, RevenueCat50:00 see cost first in Google Play surveys, but frame it as not using the app enough to justify the money. Both views point back to value.
• Can cancellation surveys be trusted?
  • Jason Cohen, WP Engine19:06 calls dropdown answers noise unless you randomize them.
  • RevenueCat still draws clear patterns from Google Play cancellation surveys, such as the 70% for cost plus low usage Jacob Eiting and David Barnard50:00. Treat survey results as a lead, then confirm with open-ended answers.
• Should you show subscribers which features are premium?
  • Michael Segal and Mark Ungerer, Skylight22:46 do not label premium features inside the product, because hardware buyers commit after months of research and do not weigh each feature.
  • Renewal rates are pretty high for both heavy and light users of the bundle. This may not fit apps where users judge value feature by feature.