Cancellation Flow Examples: A Rubric for Grading Any Cancel Flow

I've spent the last few years consulting on retention for SaaS and subscription businesses. The most consistent pattern I saw was the way teams polish how their cancellation flow examples look and never ask which design choices actually change whether a customer stays.

A cancel flow example is only useful when it shows the link between cancel reason, offer type, and measured outcome. Most example lists are screenshot galleries. This one is a four-column rubric you can run on any flow, yours or a competitor's.

Real examples show each column in action.

Key takeaways

• Grade cancel flow examples on cancel reason, offer match, and outcome, not screenshots alone.
• Offer-reason match predicts save rate more than friction level or page polish.
• A pause saves near 25% of temporary-reason cancellers, under 10% when it misfits.
• 7 of 10 flows graded (70%) show one offer regardless of the cancel reason.
• A clean, low-friction flow can still lose customers when the offer misses the reason.

What should a cancel flow example actually show you?

A cancel flow example is worth studying when it shows the connection between cancel reason, offer type, and outcome. The page design matters least, and the decision logic behind it matters most.

A screenshot tells you where the buttons sit. A useful example names three things instead. It names the offer the team deployed, the reason it was built to answer, and how often customers accepted it.

Those three inputs turn a picture into a test you can run on your own flow next week. Without them, you copy a layout and hope the result carries over.

When we built the cancel flow for a B2B fintech client, I couldn't pick an offer first. I had to know why people were leaving. Their top reason turned out to be onboarding confusion, not price. A discount would have been wrong here. It cuts revenue from customers who are already confused and still leaving.

The reason had to come first, and the offer followed from it. That sequence is what a good example makes visible.

Almost no one publishes outcome data on a named flow. That scarcity is what makes the few public numbers worth studying.

Chargebee's published case reports that Powtoon increased its save rate from 8% to 13%. The lift came after Powtoon rebuilt its cancel flow around one targeted offer.

That is the format a useful example takes. It names the change and the before-and-after numbers. With those, you can judge whether the design choice is worth borrowing.

One caution before the grades. An example read against the wrong audience fails the same way a mismatched offer does. Grade a flow built for a high-value B2B app by consumer-app standards and you get the wrong conclusion.

Take, for example, a two-week pause: it means one thing to a $9-a-month meditation app and something else entirely to a $900-a-month analytics tool.

Read every grade below in the context of the business model it was built for.

The Cancel Flow Grading Rubric

The rubric grades a cancel flow on four columns. They are the exit survey, the offer-to-reason match, the path friction, and the legal floor for online cancellation. Each column is measurable on its own, and that independence is the point:

1. Exit survey present: the flow asks why before it offers anything, with reason options written the way customers talk.
2. Offer type matched to cancel reason: the offer shown depends on the reason given, rather than one default for everyone.
3. Friction level (1-5): how many steps and obstacles sit between cancel intent and confirmation. A 1 is one clean click. A 5 is a dark pattern.
4. Compliance posture: whether the online cancel path meets the federal simple-mechanism rule and applicable state auto-renewal laws.

A flow can score high on three columns and low on the fourth. That fourth column usually decides how many customers you save. It's why "good flow" and "bad flow" labels miss the operator insight.

Column 1. Exit survey present

The first column asks whether the flow collects a cancel reason before it does anything else. This is the input every other column depends on, because you can't match an offer to a reason you never captured.

A high score is one screen and one question. The reason list is short, in the customer's own words. The subscriber sees plain options like "I'm not using it enough" or "it's too expensive," not category labels.

The wording decides how good your data is. Generic labels push people toward "Other," which tells you nothing you can route an offer against.

Why collect the reason before the offer rather than after? Because the reason is what tells you which offer to show. Run the offer first and you're guessing.

That's how a flow ends up showing, say, a 20% discount to someone who already switched tools. The discount can't win the customer back, and you've spent it for nothing. The exit survey questions guide covers how to phrase reason options so they segment cleanly.

Column 2. Offer type matched to cancel reason

The second column moves the number of saved customers the most. It asks whether the offer a customer sees depends on the reason they gave, or whether everyone gets the same thing.

Four common offer types each answer a different reason. A pause answers a temporary problem like illness or a slow season. A discount answers a price objection. A downgrade meets "I don't need all of this," while a free extension meets the subscriber still evaluating it.

Match the offer acceptance rate to the reason and you save a quarter of your cancellers, not a tenth. Mismatch it and the offer is wasted.

Picture a subscriber who selects "too busy this month" and gets shown a retention offer built for price objections. They usually decline and leave. Their problem was time, not money, and a pause would have fit where the discount didn't.

Matching the offer to the reason is the design choice most tied to saving customers in our data. It's also the column most flows fail.

Column 3. Friction level

The third column scores how hard the flow makes it to cancel, on a 1-to-5 scale. One is a single clean click. Five is a dark pattern, the kind that forces a phone call or a typed confirmation phrase.

Friction is a real lever a designer can pull, but it solves a narrow problem. A small amount keeps someone from rage-quitting on a misclick, and too much does the opposite of what operators want.

The most relevant research here comes from checkout, the closest studied analog to cancellation. Baymard Institute found that 18% of checkout abandoners walk away because the process is too long or complicated.

That number tells you what a slow decision costs.

Wrap any decision in friction and a measurable share of people give up on it. On a cancel path, that means a customer files a chargeback and never comes back. A clean cancellation you could have won back later is the better outcome.

A subscriber who hits a five-screen "are you sure" gauntlet doesn't decide to stay. They dispute the charge and leave for good. So a high friction score is not a high grade. On this rubric, low friction scores well, and we penalize the flows that pile on screens.

Column 4. Compliance posture

The fourth column asks whether the flow clears the legal floor for online cancellation. Treat it as pass-or-fail before any optimization. A flow that breaks the law exposes the company to penalties no number of saved customers can offset.

Two laws set the floor. The federal Restore Online Shoppers' Confidence Act has one rule that matters here. Anything signed up for online can be cancelled online, through a mechanism as simple as sign-up.

The state laws go further than that. Take California's Automatic Renewal Law (AB 2863, in effect July 1, 2025). It says you can cancel online "at will, and without engaging any further steps that obstruct or delay." That rules out friction.

Some 30 states now have a law of this kind.

A subscription you signed up for in two clicks should not force a phone call to cancel. One that does fails this column outright. It also fails Column 3, since a phone-only path is maximum friction.

The two columns often move together, but they grade different things. Column 3 measures the saves and chargebacks you lose, while Column 4 measures the legal risk the company takes on.

A flow can be annoying without being illegal. An illegal flow is almost always annoying too.

The offer column decides the most, and it's where our data is clearest

Of the four columns, offer-reason match moves the number of saved customers the most. Here is the evidence behind that claim, then real flows that show the column in action.

Pause offers are where the gap is widest. Match a pause to a temporary reason and roughly two to three times as many accept it as accept a discount.

A pause solves a short-term problem without locking in a lower price. Someone out sick or in a slow season doesn't need a lower price. They need a month off.

Show that customer a discount and you fix a price gripe they never raised. Worse, you teach them to expect a discount next time.

Our own data puts numbers on the gap. We show a pause to subscribers citing temporary reasons, like illness, a cost crunch, or a slow season. Roughly 25% accept it.

That figure lines up with the save rate benchmarks Recurly published in its 2025 industry report. Drawn from 67 million subscribers, that report found 25% pause instead of cancelling when offered a pause.

Show that same pause to feature-dissatisfied subscribers, the ones leaving because the product is missing something, and under 10% accept it, based on our own data. Same offer, different reason, and the share who accept moves by a factor of two and a half.

The number is about the right offer being sent only when the reason fits.

Real cancel flows, read through the rubric

The examples below come from published teardowns of each flow, not from a live audit we ran ourselves.

Sourcing matters here, because cancel UIs change and a screenshot ages. Each example names the source and its date, so you can check the current flow before you copy anything. 

Read them as illustrations of the columns, not as fixed grades.

Where the offer fits the reason

The strongest flows pick the offer to match the reason a customer just gave. A few published examples show the shape.

Userpilot's May 2026 teardown describes Slack asking why a customer is leaving before it makes any counteroffer. That order is the whole point of Column 1. The reason comes first, and the offer follows from it.

The same teardown describes Mailchimp warning annual subscribers about the data they lose on cancel, then running a four-question survey. The friction warning is honest, and the survey gives the flow something to route on, if it uses the answer.

On the consumer side, Recurly's December 2025 writeup describes Cinemark Movie Club making a pause its most visible option. For a membership people drop in a slow season, a prominent pause meets the most common reason head-on.

Recurly quotes the team crediting pause as a successful part of putting the customer first.

What these have in common is that the offer on screen has some relationship to why the customer is leaving. That relationship is the thing a screenshot can't show you and the rubric can.

Where the offer ignores the reason

The more common pattern shows one offer to everyone, whatever reason they give. The flow can be clean and compliant and still leave saves on the table.

Userpilot describes Mixpanel running a detailed exit survey and then doing nothing with the answer, no offer routed off it. That's a flow that pays Column 1's cost and keeps none of the benefit.

FunnelFox's July 2025 teardown describes Headspace offering a 50% discount when a customer cites cost. That's a clean match for the price-objection reason. The risk is the customer leaving for a non-price reason, who gets the same discount and declines it.

A discount is the right offer for one segment and the wrong offer for everyone else, shown to all at once.

FunnelFox describes Strava steering cancellers toward an annual plan at a lower per-month cost. A plan switch fits the customer who finds the monthly price steep. It does nothing for the one who got too busy to ride.

The offer is where these flows lose customers, and it doesn't show up in a screenshot. The offer looks fine on the page. It's only wrong relative to a reason the flow never asked about, or asked about and ignored.

Read every example against its business model

One caution ties the examples together. Say a $9-a-month consumer app and a $900-a-month B2B tool. They answer different reason mixes, so the same offer scores differently.

A pause fits a habit product like Headspace or Strava, where "I got too busy" drives a large share of cancellations. The customer didn't decide the product is worthless. They decided this month is full, and a pause keeps the account alive until the season passes.

A downgrade fits a B2B tool. There the common reason is "I don't use the top tier." Recurly describes Resource Guru putting plan options side by side at cancel time. The customer steps down instead of out.

Grade a flow built for one model by the other's standard and you'll draw the wrong conclusion. So run the rubric on your own flow, against your own reason mix, rather than copying the layout of a flow built for someone else's.

If you want a wall of screenshots to browse, the 17-flow Figma community file collects 115 of them. It's a useful visual reference, though it grades nothing.

What the clean-but-losing flows reveal

A flow can score high on friction, meaning a short, polished cancel path, but low on offer-reason match. Those flows look good and save fewer customers than a clunkier flow with the right offer. The lever is the match, not the polish.

Friction-first design solves a different problem than the one operators think they're solving. A clean, short cancel path lowers the chance a customer quits the flow out of annoyance.

It does nothing about the reason they want to leave.

You save the customer when the offer fits the reason, not because the page renders beautifully. A subscriber leaving over a missing feature declines a tidy downgrade screen as fast as a cluttered one.

Our own data makes the point directly. Flows with three or more steps but a well-matched offer save more customers than single-step flows with a blanket offer. The matched offer is one the flow chose from a reason it detected.

Adding friction to a mismatched offer doesn't rescue it, and stripping friction off doesn't either. The offer is either right for the reason or it isn't, and step count barely changes that.

So to fix the offer column, you have to detect the cancel reason at the moment the customer cancels. You can't route on a reason you didn't capture, and you can't capture it once they're gone.

This is the work our cancel flow builder is built to do. It runs the exit survey, reads the answer, and routes the offer to the reason in the same flow. The match happens automatically instead of by hand.

The grades above cover flows that follow the law. They leave out dark-pattern flows. Those are the ones that make you call a phone line, type a phrase, or wait days to cancel. They fail for a different reason than a mismatched offer.

Dark-pattern flows expose the company to penalties under the federal simple-mechanism rule and state auto-renewal laws. They also earn a name for being hard to quit. The rubric leaves them out because it grades flows trying to save customers honestly, not flows trying to trap them.

FAQ

What should a cancel page say?

A cancellation flow should ask why the customer is leaving, then show one offer matched to that reason. The reason question comes first, in plain language the customer would use.

Should a cancel flow have an exit survey?

Yes, because the survey answer is what lets you match the offer to the cancel reason. Without it, every offer is a guess. Keep it to one screen and one question with a few clearly worded reasons, so it adds no friction.

What offer should a cancel flow show?

Show the offer that matches the stated cancel reason. A pause fits a temporary break, a discount fits a price objection, and a downgrade fits the subscriber who needs less. A single default offer fits one of those reasons while missing the rest.

How many steps should a cancel flow have?

Two screens fit most subscription products, one for the reason and one for the matched offer and confirmation. Step count matters less than offer match, since a matched offer at three steps beats a blanket offer at one. Past three screens, each added step raises chargeback risk without saving any more customers.

‍