The cancel click is the most concentrated moment of intent in your product. Most companies answer it with a guilt-trip popup and a blanket discount.
A cancellation flow uses that moment. You learn why they're leaving, and you offer a real alternative if one exists.
How many you save depends on the product and on how you count.
I built cancellation flows at Codecademy during the run from $10M to $55M ARR (2017-2021), and I've audited many more since.
The two failures I see most: no flow at all, or every canceller gets the same 50% discount. Match the offer to the reason.
Below: the 3-step playbook, real cancellation flow examples, paste-ready copy, save rate, and dark patterns.
Cancellation Flow at a Glance
| No flow (immediate cancel) | Designed flow (ask → offer → clean exit) | |
|---|---|---|
| What happens | Click cancel, subscription ends | One-question survey, one reason-matched offer, then a one-click confirm |
| Data captured | None | Cancel reasons in the customer's own words, every week |
| Offers | None | Pause, downgrade, time-limited discount or a fix, matched to the reason |
| Time to ship | Already shipped | A basic version in days with your billing platform; 1–2 sprints custom |
| Consumer-protection risk | Low | Low if the exit stays easy; high if you hide or obstruct it |
| What to measure | Nothing | Save rate (net of re-cancels at 30/60/90 days), offer take-up, revenue saved |
The 3-Step Cancellation Flow
User clicks Cancel → 1. Ask why → 2. Reason-matched offer → 3. Clean exit
Ask why, make one offer, then let them leave.
-
1
Ask why
One question: what’s the main reason you’re cancelling? The answer picks the offer.
-
2
One matching offer
Pause, a cheaper plan, or a fix for the stated reason. One attempt, two at most.
-
3
Clean exit
If they decline, cancel immediately. Say what happens to access, when billing stops, and how to come back.
Step 1: Ask Why They're Leaving
Before you process the cancellation, ask one question: "What's the main reason you're cancelling?" The answer drives everything after it.
Write the options in your customers' words, not yours. "Insufficient feature set" isn't something a real person says. "I can't find what I need" is.
- Start with an open text field for your first 100–200 cancellations.
- Read every response and cluster them into themes.
- Turn the top 5–6 themes into multiple-choice options using the words people actually wrote. Keep an "Other" option with a text box.
A practical starter set, to replace with your own data as soon as you have it:
- Too expensive / not worth the price
- Not using it enough
- Need a break (temporary)
- Switching to something else
- Missing something I need / technical problems
- Got what I came for
Expect the first two to dominate. Churnkey's State of Retention 2025 looked at about 3 million cancellation sessions. In 2024, "Budget limitations" (32.97%) and "infrequent usage" (30.6%) were the top two.
I see the same split. As I wrote in my churn benchmarks post, subscribers don't see enough value to pay with time or money.
Put the data where people see it: a weekly dashboard for product, marketing, and leadership. A shift in reasons is an early warning.
Step 2: Show One Reason-Matched Offer
The offer has to address the stated objection. A blanket "50% off for 3 months" does nothing for someone who is too busy.
How many attempts? One, two at most. A second angle ("would a cheaper plan work?") is fine. A third kills your chances of a win-back later.
Step 3: Let Them Leave Cleanly
If they decline, cancel immediately and say what happens to access, when billing stops, and how to come back. No second guilt-trip popup.
Cancellation Flow Examples
Well-known subscription companies handle cancellation differently. Flows change often, so each example is dated. If you're copying one, check the current version first.
Pattern Table: Reason → Offer → Example
| Cancel reason | Offer that fits | Who does something like it |
|---|---|---|
| Not using it right now / need a break | Pause with a clear resume date | YouTube TV, Netflix, Zoom |
| Won't need it for a while (seasonal) | Free extension of the billing date | CLEAR |
| Too expensive | Downgrade to a cheaper tier, or a time-limited discount | Zoom (tier counteroffer), Spotify (cheaper and shared plans), Paramount+ via Amazon (2 months at $0.99) |
| Technical or product problem | Route to support or a fix before the cancel completes | Zoom (by account type) |
| Got what I came for | Congratulate, offer a referral or alumni option | Common in education and fitness |
Real Cancellation Flows: What Each Company Offers
| Company | Primary save offer | Reason-specific? | Pause option? | What to copy | Checked |
|---|---|---|---|---|---|
| CLEAR | Free billing-date extension (2–3 months observed) | No, one offer to everyone in the self-serve flow | Effectively (the extension works like a pause) | A zero-cost offer for "won't use it for a while" | May 15, 2026 |
| YouTube TV | Pause, 4 weeks to 6 months, user-selected | Not documented in the current self-serve flow | Yes | Let the customer pick the pause length | May 15, 2026 |
| Zoom | Reason survey, then a counteroffer (lower tier or pause up to 3 months) | Yes | Yes (paid plans, up to 3 months) | Survey first, then route by reason and plan | May 15, 2026 |
| Netflix | Pause (1 month at a time, up to 3 months total) | No | Yes (not available on every plan or billing method) | Pause inside the cancel flow, with a fixed cap | May 15, 2026 |
| Spotify | A loss-framing screen, then cheaper or shared plans (Duo, Family, Student) | No | No formal pause | Plan alternatives as the save offer instead of a discount | Screenshots captured Jul 2023–Jun 2024 (dated by the prices shown); help doc May 15, 2026 |
| The New York Times | Not described in the help doc | Not documented | Not documented | Self-serve online cancel offered alongside phone and chat | Help doc checked Nov 14, 2026 |
| Paramount+ (sold through Amazon Prime Video Channels) | $0.99 a month for 2 months, then $12.99 | Not visible in the screenshot | Not shown | A deep, time-limited discount that states the price it returns to | Screenshot captured around late Nov 2025 |
The best flows match an offer to a reason and still make leaving easy. Netflix pauses without routing by reason, because "I've watched what I wanted" isn't fixable in the flow.
What the Screens Actually Look Like
These screenshots come from my research library. Each one is dated by what's visible on screen.
YouTube TV: reasons first, then a pause. The cancel flow asks for a reason. Choosing "Cost" opens two alternatives: share the plan with up to 4 household members, or pause.

The pause screen lets the subscriber pick the length and shows the date billing resumes. The current help doc describes pause lengths from 4 weeks to 6 months.

The insider subscription playbook
Get new guides and teardowns by email
Real companies, real numbers, strategies you can steal. From the operator who grew Codecademy from $10M to $55M ARR.
Copy: reason-first routing, a customer-controlled pause, and a stated resume date. Don't copy blindly: 6 months suits seasonal sports. For most products, that's long enough to forget you.
Spotify: loss framing, then cheaper plans. The help doc describes a short cancel path. The captured flow added what you lose on the free tier, then shared and discounted plans.


Copy: a cheaper or shared plan beats a discount for "too expensive," because the subscriber stays at full price per seat. Don't copy: loss framing on every reason. It feels like pressure on a break.
Paramount+ (through Amazon Prime Video Channels): the deep, time-limited discount.

Copy: the screen states the price the subscriber returns to and the date. Watch: a 92% discount is a strong save that's likely to churn when it ends. Measure it at 90 days.
CLEAR and Zoom, briefly. When I cancelled CLEAR, the flow pushed my next billing date out by a couple of months at no charge. Don't let that repeat forever, or it becomes a pricing loophole. Zoom follows a reason survey with a lower tier or a pause of up to 3 months on paid plans.
Everyone who leaves goes into your win-back sequence. The original thinking is in my deep dive on cancellation flow best practices.
Offers That Work, Especially Pause
The Offer Playbook
Each cancel reason gets one offer.
Too expensive
A cheaper plan, an annual plan at a lower monthly equivalent, or a discount limited to 2–3 months.
Not using it enough
Pause for 1–3 billing cycles, plus one specific next step.
Need a break
Pause, with a clear date billing resumes.
Switching
Ask what they’re switching to. Treat it as intelligence, not a save.
Missing something, or a technical problem
Route to support or a fix, and offer a pause until it’s resolved.
Got what they came for
Congratulate them. Offer a referral reward or a cheaper alumni plan.
| Cancel reason | Best offer | Why it works |
|---|---|---|
| Too expensive | Downgrade to a cheaper plan, an annual plan at a lower monthly equivalent, or a time-limited discount (2–3 months) | Closes the value gap without permanently resetting price |
| Not using it enough | Pause for 1–3 billing cycles, plus help getting value (one specific next step) | Removes the payment while keeping the door open |
| Need a break | Pause with a clear resume date | It's exactly what they asked for |
| Switching to something else | Ask what they're switching to and why. Treat it as intelligence, not a save. | Low save rate, high roadmap value |
| Missing something / technical problem | Route to support or a fix; offer a pause until it's resolved | Shows you're listening and creates a win-back trigger |
| Got what I came for | Congratulate them; offer a referral reward or a cheaper "alumni" plan | Happy leavers become referrers. Don't fight them. |
For "too expensive," an annual plan at a real per-month saving can rescue people who can't justify the monthly rate.
Why Pause Is the Most Underused Offer
At Codecademy, the most common cancel reason was "life got busy." They meant to keep learning and couldn't right now. A discount did nothing. The objection was time, not price. A pause matched it.
In Recurly's 2025 State of Subscriptions, businesses that offered a pause saw 25% of subscribers choose it instead of cancelling. Subscribers who resumed after pausing generated more than $200M.
Recurly's write-up found 55% of consumers rank pause among the features they want most, while only 37% say their services offer it. Paused subscribers restart with no extra steps. Cancelled subscribers have to re-subscribe.
The pause works when it matches the reason. It doesn't fix content exhaustion or a product that isn't delivering. Those subscribers unpause and cancel later.
How Long Should a Pause Last?
There's no universal standard. The examples above range from Netflix's 1 month at a time (3 months max) to YouTube TV's customer-chosen 4 weeks to 6 months.
- Tie it to your billing period. One to three billing cycles is a sensible starting range for monthly plans.
- Tie it to the reason. Seasonal products (sports, tax, travel, school terms) can justify a pause that covers the off-season.
- Cap it. A pause long enough to forget about is really a delayed cancellation.
- Be explicit. Show the exact date billing resumes, send a reminder a few days before, and make extending or cancelling from the reminder one click.
If most paused subscribers cancel the day billing resumes, the pause is too long, or you offered it for the wrong reasons.
When a Discount Is Right
Use a discount only for a real "too expensive" reason, and only for a limited time. A permanent discount is contraction in net revenue retention forever. Check if those subscribers stay after it ends.
Membership and Media Variants
For $50M+ membership and media businesses, use a cheaper tier (digital-only instead of print + digital), a billing-cycle change, a pause around a season, or a conversation for high-value members. On a price-increase cancel, acknowledge the increase and offer the cheaper tier.
Paste-Ready Cancellation Flow Copy
Swap in your product, plans, and prices. Keep the "cancel" button visible on every screen.
Survey Screen
What's the main reason you're cancelling?
One question. Your answer goes straight to our product team.
- It's too expensive for me right now
- I'm not using it enough
- I need a break. I'll be back.
- I'm switching to something else
- Something's missing or not working
- I got what I came for
- Other: [text box]
[Continue] [Skip and cancel]
Offer Screens, by Reason
Too expensive (downgrade first, discount second):
Want to keep [Product] for less?
Switch to [Cheaper plan] for $[X]/month. You keep [top two features]. You lose [what's removed].
[Switch to [Cheaper plan]] [No thanks, cancel]
Take [N]% off for the next [2–3] months
After that, it's $[price]/month. We'll email you before the price changes.
[Apply discount] [No thanks, cancel]
Not using it enough:
Not getting to it? Pause instead.
Pause for 1, 2 or 3 months. You won't be charged, your [progress / library / settings] stays put, and billing restarts on [date]. We'll email you a week before.
Or try this first: [one specific next step, such as a 10-minute lesson or a saved playlist].
[Pause my subscription] [No thanks, cancel]
Need a break:
Take a break and keep your account.
Choose how long: [1 month] [2 months] [3 months]. Billing restarts on [date]. You can end the pause early or cancel any time.
[Start my pause] [No thanks, cancel]
Switching to something else:
Mind telling us where you're going?
What are you switching to, and what does it do better? [text box]
[Send and cancel] [Just cancel]
Something's missing or not working:
Sorry about that. Can we fix it first?
Tell us what went wrong and [team name] will reply within [timeframe]. If you'd rather not wait, you can pause until it's fixed.
[Get help] [Pause instead] [Cancel]
Got what I came for:
Congrats on [the outcome].
If you know someone who'd get the same value, here's a link that gives them [reward].
[Share my link] [Cancel subscription]
Cancellation Confirmation Email
Subject: Your [Product] subscription is cancelled
Hi [First name],
This confirms we've cancelled your [Plan] subscription. You won't be charged again.
You'll keep access until [date], the end of the period you've paid for. Your [progress / saved items / account] will be here if you come back, and restarting takes one click: [Reactivate].
If you didn't mean to cancel, or have a question, just reply to this email.
Thanks for being a subscriber.
[Name], [Product]
Pause-Ending Reminder Email
Send it 7 days before billing restarts.
Subject: Your [Product] pause ends on [date]
Hi [First name],
Your pause ends on [date]. Your [Plan] subscription restarts that day at $[price]/[month or year].
Since you paused: [one line on what's new].
Need more time? [Extend my pause] (up to [maximum]). Not coming back? [Cancel] takes one click.
Otherwise, you don't need to do anything.
[Name], [Product]
Some state auto-renewal laws set their own rules for renewal and cancellation notices. Check with counsel for the markets you sell in.
Build vs Buy
Start with your billing platform. Stripe's portal can collect cancellation reasons and show a retention offer, and retention rules can target offers by reason. Paddle, Chargebee, and Recharge have equivalents. A basic flow can be live in days.
Build it yourself when the offer depends on product data like usage, progress, or content.
Buy a dedicated tool (Churnkey and others) when you need faster tests, offer logic your platform lacks, cancel volume where a few points of save rate is real money, or nobody owns retention engineering.
What to Measure
Save Rate: Define It Before You Report It
Save Rate = Cancel intents that did not end in cancellation ÷ Cancel intents that started the flow × 100
Define three things up front:
- The window. Same day, or still active after 30/60/90 days.
- What counts as a cancel intent. Exclude bounces and accidental clicks the same way every time.
- Voluntary only. Exclude failed-payment cancellations. Those belong to failed payment recovery.
Tools define this differently. Churnkey counts abandoned sessions as saved and excludes bounced sessions, and reports an average save rate of 34%. That's their own base. Don't compare it with a number calculated differently.
The Number That Matters Most: Net Save Rate at 90 Days
A flow can look great on day one and collapse by month three. Someone who takes 50% off and cancels when it ends wasn't saved. You delayed the cancel at a lower ARPU. Track "saved" subscribers still paying at 30, 60 and 90 days.
Also Track
- Save rate by reason: which offers actually work
- Offer take-up: especially pause take-up, and what happens when pauses end
- Revenue saved, net of discount cost
- Survey completion rate: a low rate usually means the question is badly placed or too long
What's a Good Save Rate?
There's no single good number. Ecommerce, consumer apps, and media don't share one. Use the 90-day net save rate, by reason, and trend it. If you test the flow, check the MDE calculator first. Cancel flows often have less traffic than you'd think. Confirm the result with the statistical significance calculator.
Mistakes and Dark Patterns
Dark Patterns
Hidden cancel buttons, mandatory calls, chained screens, fake urgency, and vague pause terms can lower cancellations for a while. They also destroy trust and draw regulators.
The regulatory picture (US, as of November 2026, not legal advice):
- The FTC's 2024 "click-to-cancel" amendments were vacated by the Eighth Circuit on July 8, 2025.
- On February 12, 2026, the FTC restored the older Negative Option Rule.
- On March 13, 2026, the FTC published an advance notice of proposed rulemaking asking whether to adopt new rules on online subscriptions and cancellation.
- Meanwhile, the Restore Online Shoppers' Confidence Act (ROSCA) still requires a simple mechanism to stop recurring charges for online subscriptions, and the FTC continues to bring cases.
- State auto-renewal laws add their own requirements, and other markets (UK, EU, Australia) have their own consumer-protection rules.
Check with counsel for the markets you sell in.
Practical design principles that keep you safe and don't cost you saves:
- The cancel path is easy to find and can be finished online, in the same channel the person signed up in.
- One survey question and at most one or two offers, with a visible "No thanks, cancel" on every screen.
- Pause and discount terms state exactly when billing resumes and at what price.
- Confirm the cancellation immediately, by email.
A good cancellation flow makes it easy to cancel and easy to stay.
Generic Discounts
Showing everyone 50% off says you weren't listening, and it trains subscribers to threaten cancellation for a deal.
Ignoring the Data
Cancel reasons are free product research. Review them weekly with product.
Skipping Mobile
If people subscribe through the App Store or Google Play, they cancel there and skip your flow. On iOS, listen for App Store Server Notifications. DID_CHANGE_RENEWAL_STATUS fires when a user turns off auto-renew, and EXPIRED with subtype VOLUNTARY fires when the subscription lapses. Use that for a win-back, or a prompt the next time they open the app.
When NOT to Build a Cancellation Flow
Your voluntary churn is already low. A good save rate on a tiny base doesn't add up. Spend the time on pricing, onboarding, or the paywall.
You have very few cancellations a month. You won't get enough answers to trust the categories. Ship the survey, collect for a few months, then build offers.
Most of your subscribers cancel through the app stores. An in-app flow only catches part of the intent. Build win-back first.
You haven't fixed involuntary churn. If a big share of churn is failed payments, failed payment recovery and a dunning process are cheaper. Do them first.
FAQ
What is a cancellation flow?
A cancellation flow is the set of screens between "Cancel" and a completed cancellation. A good one asks why the subscriber is leaving, offers one reason-matched alternative (pause, downgrade or a time-limited discount), and then lets them leave in one clean step if they still want to.
What should a subscription cancel flow include without being a dark pattern?
An exit survey, one reason-specific offer (pause, downgrade or discount), and a clear path to finish the cancellation. It shouldn't hide the cancel button, add surprise steps, use fake urgency or make pause terms unclear. This is practical design guidance, not legal advice. Check the rules in your market.
What's a good save rate for a cancellation flow?
Save rate = cancel intents that didn't end in cancellation ÷ cancel intents that started the flow, over a defined window, excluding failed-payment cancels. There's no universal good number. Ecommerce, consumer apps and media memberships differ, and tools define save rate differently. Track the net save rate at 90 days, segment by reason, and trend it.
Should we offer a pause, and for how long?
Yes, for "too busy," "need a break" and "not using it right now" reasons. In Recurly's 2025 State of Subscriptions, businesses offering a pause saw 25% of subscribers pause instead of cancelling. Set the length to a meaningful break, often one to three billing cycles or a seasonal window. Show exactly when billing resumes, and send a reminder before it does.
What are good cancellation flow examples to copy?
YouTube TV asks for a reason, then offers plan sharing or a pause with a stated resume date. Spotify shows what you lose, then offers cheaper shared plans. Paramount+ (through Amazon Channels) offered $0.99 a month for 2 months with the return price stated. Netflix offers a capped pause. The screenshots and the dates they were captured are in the examples section above.
What should a cancellation confirmation email say?
That the subscription is cancelled, that they won't be charged again, when access ends, what happens to their data, and how to come back in one click. Send it immediately. Don't add another save offer to it. There's a paste-ready version above.
Should every cancelling subscriber get a discount?
No. Offer discounts only to the "too expensive" segment, and make them time-limited. For every other reason, match the offer to the problem. Generic discounts teach people to threaten cancellation to get a deal.
Does this work for membership or B2B subscriptions?
Yes, but the offers change. Membership and media businesses lean on cheaper tiers, billing-cycle changes, pauses and member-services conversations. B2B products may need seat reductions, contract flexibility or a call with the account team rather than a coupon.
How do I handle subscribers who cancel and come back repeatedly?
Track previous cancel attempts per subscriber. On a repeat attempt, show a different offer or go straight to a clean exit. The value of a save offer drops each time it's used by the same person.
How is a cancellation flow different from win-back?
The cancellation flow happens during cancellation. Win-back starts after the customer has left, using their cancel reason to personalize the follow-up, for example at 7, 30 and 90 days. See win-back campaigns.
What to Do Next
If you don't have a cancellation flow, I'd go in this order:
- Add a one-question survey to your cancel button. It takes hours, and the data is valuable immediately.
- Add a pause option if your billing platform supports it.
- Build reason-specific offers once you have a few hundred responses. Map your top three reasons to targeted offers.
- Measure net save rate at 30/60/90 days by reason. Kill offers that don't hold up.
Some subscribers should leave. Offer a small adjustment to the ones who'd stay. Pair the flow with win-back, dunning for involuntary churn, and your retention strategy. The churn rate calculator shows what a few points of churn are worth over a year.
Get new guides and teardowns by email. Browse the archive or subscribe below.
If you want a cancellation flow designed or audited (survey, offers, pause, measurement), including for a membership or media team, I do a free 30-minute call. No pitch.
