Cancellation flow benchmarks:
which saves last.
A cancellation flow is judged by what happens after the save. We followed subscribers who tried to cancel in the stores we run and audit for two months after the save. We read in their subscription apps which saves lasted and what each one brought in.
Cancellation flow benchmarks, in short.
of people who try to cancel take a save at top stores. The typical store is at 21%.
are still subscribed a week later at top stores, the true save rate. The typical store is at 9%.
of saves on monthly plans are still subscribed two months later at top stores. At the typical store it is 26%.
comes in for every $1 a month saved at top stores. At the typical store it is 44¢.
revenue per save comes from an offer than from a click on a benefits page. That is after the discount.
Top cancellation flows save up to 44%.
At top stores 24% to 44% of people who try to cancel take a save. At the typical store it is 21%. A week later 10% to 13% are still subscribed at top stores, and 9% at the typical store.
| Top stores | Typical store | |
|---|---|---|
| Took a save (platform save rate) | 24% to 44% | 21% |
| Still subscribed after 7 days (true save rate) | 10% to 13% | 9% |
12 to 25 October 2026, each subscriber counted once. An app’s own screen counts each visit and can read lower. Top stores are the top quarter, up to the strongest store. Each number can come from a different store.
The gap between the two rows is people who click to stay and cancel anyway, often minutes later. The app counts them at the click. The store at the top of the first row is not the store at the top of the second, so a report needs both.
Benchmarks quoted online, often 20% to 35%, are the app’s count too, so set them against the first row.
The strongest flow keeps nearly half of its saves for two months.
On monthly plans, 32% to 47% of saves are still subscribed two months later at top stores. At the typical store it is 26%. We count monthly plans here because a renewal falls due inside the two months.
| Of every 100 saves on monthly plans | Top stores | Typical store |
|---|---|---|
| Still subscribed after 7 days | 58 to 59 | 47 |
| Still subscribed after 30 days | 43 to 54 | 36 |
| Still subscribed after 60 days | 32 to 47 | 26 |
Saves from November 2025 to September 2026, on plans of up to five weeks. Still subscribed means no cancellation yet, and a pause counts. Top stores are the top quarter, up to the strongest store. Each number can come from a different store.
The largest single loss comes in the same visit: about four saves in ten are undone before the subscriber leaves the page. That is where to look first.
Top stores collect up to $1.47 for every $1 a month saved.
At top stores every $1 a month a flow saves has brought in 90 cents to $1.47. At the typical store it is 44 cents.
| Top stores | Typical store | |
|---|---|---|
| Revenue for every $1 a month saved | 90¢ to $1.47 | 44¢ |
| More orders from each saved subscription | One to two | About half |
The app’s own save report, saves from November 2025 to April 2026, read in November 2026, so every save has had at least six months. Stores on longer plans read higher.
This is the number to judge a flow on, because it is the one that reaches the bank.
Offers bring in the most money, pauses last the longest.
Two in three saves are a click on a benefits page. Offers are 14% of saves, skips 8% and pauses 2%.
An offer
Brings in about a third more revenue per save than a click, even after the discount. It also lasts longer at first, and by two months the two are level.
A pause
Lasts the longest. Nobody pays during a pause, though, so the money comes later or not at all.
A click on a benefits page
The most common save, at no cost in discount. It outlasts a skip.
A skip
The shortest-lived save. An offer outlasted it at every store that had both. On Skio a discount brought in more than three times the orders of a skip. Too much product is a real reason to leave, so keep it, and count it once the next order is paid.
The first day is the easiest save.
At top stores 15% to 18% of people who try to cancel on their first day are still subscribed a week later. At the typical store it is 12%, and every later stage reads well below that.
The reason matters too. People with too much product or a price problem are three to four times as likely to stay. The hardest to keep no longer need it or see no results. A flow can change a delivery date or a price. It cannot make a product work.
A higher save rate is not always more people kept.
The app’s count can rise without keeping more people. Between mid September and late October it rose from 17% to 21% at the typical store. The share still subscribed a week later held at about 8% to 9%.
A new page or a new step can add clicks without adding subscribers. In one store, on one flow, the two numbers usually move together. So the app’s count is a fair way to compare a store with itself.
Some apps count a closed page as a save.
On Recharge, someone who closes the page without answering counts as saved, and that is about half of its saves at the typical store. Many of those people do stay, but they have not chosen to.
So a save count does not compare directly from one app to another. To compare across apps, use the share still subscribed a week later.
Work out your true save rate from the cancellation log.
Your true save rate is the share of people who try to cancel and are still subscribed 7 days later. It comes from your subscription app’s cancellation export.
Get the export
Export your app’s cancellation log as far back as you can. Each app names it differently. It holds customer emails, so keep it private.
Check each save
For each save, look for a cancellation on the same subscription in the next 7 days. If there is none, they are still subscribed.
Do the sum
Divide the saves still there by everyone who tried to cancel. That is your true save rate. Do it again at 30 and 60 days.
Leave out new saves
A save from yesterday has not had time to come undone. Only count saves that are old enough.
Count closed pages apart
If your app counts a closed page as a save, work out the rate with those people and without them.
Cancellation flow best practices, from the numbers.
Report both save rates
Put the true save rate next to the app’s number in every report, so a jump on the screen cannot mislead.
Ask why first
Too much product and price are the reasons a flow can answer. Offer a longer gap between orders or a smaller pack before money off.
Count offers in money
An offer brings in more than a click. Check the revenue that follows it, along with the save.
Move fast on new subscribers
The first day is the best chance a flow gets. Give those people their own screen.
For the full plan, read our subscription churn strategy. For where subscribers leave, see our subscription churn benchmarks. New to these terms? Try our subscription marketing glossary.
Proof on named accounts.
save rate, up from 4.17%, by the app’s count, Zapply UK. Read the case study
subscription churn, Frøya Organics. Read the case study
Real client accounts, named and linked. All case studies.
How we measured.
This study uses the stores YOCTO runs and audits. Every figure is anonymous and pooled; no store can be told apart from another.
The data
The tables read each store’s cancellation log from Loop, November 2025 to November 2026, the export that records each step of the flow. Recharge’s cancellation activity with its cancelled subscriptions covers the same year and is read apart, because it counts a closed page as a save. A store counts from the first save in its log to the last, because a visit outside that span could not have been saved.
A visit and a save
A visit is one subscription’s run of steps in the customer portal, under 30 minutes apart. A save is its run of saved entries. A save lasts until a cancellation from any source lands on that subscription.
The numbers
Typical store is the median, with each store counted once. Top stores run from the 75th percentile to the highest store in that row, and each number can come from a different store.
Periods
The first table covers 12 to 25 October 2026, the last two full weeks with 7 days to follow, each subscriber counted once. Counted by visit, as an app’s screen does, over 5 to 25 October, top stores read 26% to 38%. The 60-day table follows the same people, up to 4 September 2026, on plans of up to five weeks, so a renewal falls due inside the 60 days; across every plan length it reads 35% to 49% at top stores and 30% at the typical store. Over that span the typical store read 16% by the app’s count and 8% after 7 days, the figures on our churn benchmarks. Time subscribed and reason use every visit up to 27 October. Reasons count only people who reached the reason step.
What a save is worth
Loop’s own Saves screen: further orders per saved subscription, and additional revenue against saved monthly revenue, for saves from November 2025 to April 2026, read in November 2026. Revenue leaves out shipping, tax and refunds.
Limits
There is no holdout: some saved subscribers would have stayed anyway, and a cancelled one can come back, which the log does not show. Still subscribed means no cancellation yet, and a pause counts. Of the saves still there at 60 days whose later log rows we could read, about six in ten had placed another order by then. The stores lean to consumables and all hired or auditioned a retention agency. Flows changed during the year. Stores differ widely, so these numbers are a guide, not a promise. In the cuts by time subscribed and by reason, one owner contributes several of the stores. We could not follow saves by date on Skio, so it appears only on orders after a save.
Questions about cancellation flows.
What is a good save rate for a cancellation flow?
Top stores save 24% to 44% of people who try to cancel, and the typical store saves 21%. A week later 10% to 13% are still subscribed at top stores, and 9% at the typical store.
What is a true save rate?
It is the share of people who try to cancel and are still subscribed 7 days later. The platform save rate counts the click. The true save rate waits a week and checks.
Do saved subscribers stay?
On monthly plans, 32% to 47% of saves are still subscribed two months later at top stores. At the typical store it is 26%. The largest single loss comes in the same visit, when someone clicks to stay and then cancels.
Should a subscription cancellation flow offer a discount?
Yes, after asking why the subscriber is leaving. A save from an offer brings in about a third more revenue than a click on a benefits page, even after the discount.
Is a skip a save?
Only once the next order is paid. A skip is the shortest-lived save in the study. Too much product is a real reason to leave, so keep the option and count it when the order comes in.
YOCTO is the Klaviyo agency for DTC and subscription brands. We run customer retention, subscription marketing, email marketing, deliverability and SMS, and we start with a free Klaviyo audit.
Cite this research.
Free to cite with attribution to YOCTO, under a CC BY 4.0 license. More benchmarks are on our subscription ecommerce statistics. All our studies are on the research page. Next refresh: January 2027.
Kapernaros, G. (2026). Cancellation Flow Benchmarks 2026: How Many Saved Subscribers Stay. YOCTO. https://yocto.agency/cancellation-flow-benchmarks/
See which of
your saves last.
We read your cancellation log against these benchmarks. Then we show you which saves hold and what to change first.
