The cancel flow that saves customers and tells you why they leave

At 6% monthly churn, half your customers are gone within a year. And most of them leave without a word. They click Cancel, the subscription ends, and all you have is a line in Stripe that says it happened, not why.

A cancel flow changes that. It is a short screen between the Cancel button and the actual cancellation. It asks one question, makes an offer that fits the answer, and lets the person go cleanly if the offer doesn’t fit. Some customers stay. All of them tell you something. This guide shows how to build one, with texts you can copy.

Why most churn is silent

A customer who is unhappy rarely writes to support first. They stop logging in for a while, then they cancel. If your cancel button goes straight to “Your subscription has been cancelled”, you lose the customer and the lesson at the same time.

Small monthly losses add up faster than they look. At 6% monthly churn you keep 94% of your customers each month. After a year that is 0.94 multiplied by itself twelve times:

0.94^12 ≈ 0.48

So you keep about 48% of today’s customers. Roughly every second customer is gone after 12 months, and you have to win new ones just to stand still.

New ones are not cheap. A Harvard Business Review article from 2014 puts it plainly: acquiring a new customer is 5 to 25 times more expensive than keeping one, depending on which study you believe. The same article cites research by Frederick Reichheld of Bain & Company: a 5% increase in customer retention raises profits by 25% to 95%. These numbers come from many industries, not from SaaS alone, so read them as a direction, not a forecast. The direction is clear, though: a customer you keep is worth more than one you have to find.

Ask one question before they go

One question. Not a survey. Someone who is leaving will not fill in ten fields, and they shouldn’t have to.

Keep the answers to three or four that cover the usual reasons, plus “Something else” with a short text field. For most small SaaS products and apps, these three cover a lot:

Before you go: what's the main reason you're cancelling?

( ) It was too hard to set up
( ) It's too expensive for what I get
( ) It's missing something I need, or I don't use it enough
( ) Something else: [__________]

Anything we should know? (optional) [__________]

[Continue]   [Skip and cancel]

Three things make this question work:

• It is optional. “Skip and cancel” is right there. A forced question annoys people and gives you made-up answers.
• It asks for the main reason. One pick, not “select all that apply”. You want to know what tipped it.
• It leaves room for their words. The optional text field is where the useful sentences come from: “I couldn’t connect my calendar”, “my client stopped paying me”. Read every one.

Change the answers to fit your product. If you sell to teams, “My team moved to another tool” may belong in the list. Just keep it short.

Three answers, three offers

Each answer gets its own offer. That is the whole idea. A discount for someone who couldn’t set the product up doesn’t fix their problem; it only makes a product they can’t use a bit cheaper. Match the offer to the reason.

“Too hard to set up”: offer help

These customers wanted the result and got stuck on the way. Offer to get them there: a short call, a done-for-you import, or a person who replies today.

Let us set it up with you

Most people who tell us this were one step away. Book 15 minutes with [Your name] and we'll get [Product] working with your [data/site/account], together.

[Book a 15-minute call]   [No thanks, cancel]

“Too expensive”: offer a lower price, a cheaper plan or a pause

Price is sometimes about the price and sometimes about timing. Give one clear way to pay less: a discount for a few months, a smaller plan, or a pause with no charge while it lasts. A pause works well for people who say “not right now”.

Stay for less

Keep [Product] for 20% off the next 3 months. Or switch to [Smaller plan] for [price] a month and keep your [projects/data]. Or pause for up to 3 months and pay nothing in the meantime.

[Take 20% off]   [Switch plan]   [Pause]   [No thanks, cancel]

The 20% for three months is an example, not a rule. Pick what you can afford, and say exactly what happens after: the date the discount ends and the price from then on.

“Missing something”: show the roadmap or the workaround

Some customers need a feature you don’t have. Others have it but never found it, which often hides behind “I don’t use it enough”. If the feature is on your list, say so honestly. If there is a way to do it today, show it.

Tell us what's missing

What did you need [Product] to do? [__________]

If it's already on our list, we'll let you know when it ships. If there's a way to do it today, we'll send you the steps within a day.

[Send and stay]   [Send and cancel]

Notice the second button. They can tell you what’s missing and still leave. That answer is worth more than a customer kept by force.

After twenty answers, fix the biggest one

One answer is an anecdote. Twenty answers are a pattern. Say twelve of twenty people pick “too hard to set up”. Then your problem isn’t price, and no discount will fix it. Fix setup first: the first-run screens, the import, the one step where people get stuck. Then look at the next biggest reason.

The offers save some customers this month. The pattern saves customers you haven’t signed yet.

Let the rest go well

Some people will leave anyway. Let them. How someone leaves decides whether they come back, recommend you, or write about you.

• No guilt trip. No “Are you sure you want to abandon your team?” and no sad mascot.
• No dark patterns. No hidden button, no grey “cancel” link next to a bright “stay” button, no five-screen maze, no “call us to cancel”.
• Cancelling stays one click away. From the question and from every offer, one click finishes the cancellation.
• Say what happens next. Until when they keep access, how long their data stays saved, how to export it.

This is also the safe choice. In the US, the FTC’s “click to cancel” rule was struck down by a federal appeals court in July 2025, before it took effect. That didn’t make hard-to-cancel subscriptions a good idea. Several US states have their own cancellation rules, and the EU is pushing in the same direction. An easy cancel is fair to customers and keeps you out of trouble. This isn’t legal advice: check the rules where your customers live.

Win them back after 90 days

The people who left for a reason you’ve since fixed are your warmest leads. They already know the product, they told you exactly what was wrong, and now it isn’t wrong anymore.

About ninety days after they cancel is a good moment. That gives you time to ship the fix, and they still remember you. Mail only the people whose reason you fixed, mail them once, and write like a person, not like a campaign.

Subject: You said [Product] was too hard to set up. We fixed it.

Hi [First name],

when you cancelled [Product] in [month], you told us setting it up was too hard. You weren't the only one, so we changed it:

- [The fix in one line, e.g. "Import from [tool] in one click"]
- [A second fix, if there is one]

Getting started now takes [about five minutes]. [Your projects are still saved.]

If you'd like to give it another go:
[Reactivate link]

And if it's still not for you, no reply needed. I won't email you about this again.

[Your name], [Product]

What makes it work: it names their reason in their words, shows the fix, and makes coming back one click. Don’t send it to people who left over something you haven’t fixed. It reads as if you weren’t listening.

Catch them before the button

The best save happens before anyone opens the cancel flow. Most customers show signs weeks before they cancel. Three are worth watching:

And some churn isn’t a choice at all. A customer whose card was declined didn’t decide to go. That kind needs retries and a card-update email, not a cancel flow: Failed payments: the churn you can win back.

What to measure

Check these once a month:

About save rate: there is no independent study of what a good one is. Vendors who sell cancel flows report that good flows save roughly 15–30% of the people who start them. Read that as their own range, not a benchmark. The definition above is Churnkey’s, and it counts someone who closes the flow without cancelling as saved, which is why the third row matters.

Write down this month’s numbers before you change anything. And read the answers, not only the counts. Every cancel tells you what’s broken.