Learn why subscribers cancel, and which saves last.

An honest read on why subscribers cancel, and tests that show which saves still leave a paying subscriber 90 days later.

Start with a retention review How it works

Cancellation reasonsLast quarter

Reason	Share	Save to test
Still have plenty	41%	Skip or delay
Away or moving	18%	Pause
Price	15%	An offer
Not using it	14%	Swap the product
Something else	12%	Ask

Why Care Kit subscribers cancel, grouped by reason, with the save worth testing for each.

Sound familiar?

The cancel survey says “too expensive” for most of them, and you don’t believe it.
A bigger discount gets accepted more often, and nobody checks who’s still paying.
The reasons sit in surveys, support tickets and reviews that never meet.

A save that’s accepted and canceled a month later cost you the discount and kept nothing.

What changes.

TodayWith the engine

Reasons: A dropdown, taken at its wordSurvey, tickets and orders read together
Offers: The biggest discount for everyoneTested, reason by reason
Judging a save: The offer was acceptedStill paying at 90 days
Survey changes: A new option quietly reshuffles the dataReason codes versioned, so results stay comparable

How it works. 6 steps, 4 stages.

AI2
Rules and automation2
Your team2
Approval

1Collect
1. Rules and automation
  Capture the reason and the context
  The cancel survey, support contacts and recent orders, joined on one subscriber.
  - Recharge
  - Gorgias
2Group
1. AI
  Group the reasons
  Free-text answers become reasons your team can name, with examples kept beside each.
2. Your teamApproval
  Agree the reason codes
  Your team approves the categories, so the data stays comparable when the survey changes.
3Test
1. Rules and automation
  Randomize the offer
  Each reason gets a randomly assigned save: a skip, a swap, a discount or nothing.
  - Recharge
  - Stay
4Decide
1. AI
  Read what lasted
  Counts the subscribers still paying at 90 days, and what each save really cost.
2. Your teamApproval
  Choose what stays in the flow
  Keep, change or drop each save, with a date to check again.

Each read sets the next test

What you get. Real work, every week.

Save tests · Care KitExtra payers per 100 offers

Offer a skipTo “still have plenty”+11Keep
10% offTo price reasons+4Keep
20% offTo price reasons+5Drop

Paying at 90 days, against a no-offer group

Judged at 90 days. A bigger discount earns its place only if more subscribers are still paying.

Test planCare Kit cancel flow

Which offer keeps a subscriber paying?

Groups: A skip, 10% off, 20% off, no offer
Measured: Still paying at 90 days, after discounts
Reasons: Coded the same way throughout
Decision: Keep an offer only if it pays for itself

Every save is a test with a no-offer group, agreed before the flow changes.

How you’ll know it’s working.

Subscribers still paying at 90 days
The number a save should be judged on, which acceptance alone can’t show.
Cost per saved subscriber
Discounts and gifts, divided by the subscribers still paying who wouldn’t otherwise have been.
Cancels with a usable reason
The reasons only help if most cancels leave one your team can act on.

Baselines agreed before we build.

Works with what you have.

Commerce

Recharge
Stay
Shopify

Email & SMS

Klaviyo

Support

Gorgias

We configure first: Recharge and Stay both have cancel flows with reasons and offers. We use them to ask the questions, then fix how the results are read.
We build when: Reasons sit in several systems, or you want randomized offers read at 90 days against a no-offer group.

What a first build includes.

Built in your accounts, around your team. Scope, fee and baseline agreed before we start.

1Design
2Build
3Run it, or hand it over

Start with a review Scope this build

We build

One set of reason codes across survey, tickets and orders
A cancel flow with randomized offers
A no-offer group in every test
A 90-day readout of what each save kept

We start from

Cancellation and survey data
Support contacts from subscribers
Subscription and order history

From your team

Owner: Head of retention or lifecycle
Involves: Lifecycle
Customer service
Finance

Questions teams ask.

Isn’t this the same as subscriber control?

They’re neighbors. Subscriber control makes skipping, delaying and swapping easy. This job learns why people cancel anyway, and which saves keep them paying. We measure the two separately, because a skip that only delays a cancel isn’t a save.

Why not just offer a bigger discount?

ProsperStack, a cancel-flow vendor, reported randomized tests in 2026 where a 20% offer roughly doubled acceptance against a 10% one. It added only about one more paying subscriber per 100 offers by day 90. The vendor didn’t disclose sample sizes, so treat it as a signal. Acceptance is easy to count, and still paying at 90 days is what pays.

Can we trust the reasons people pick?

Treat them as a starting point. Wording changes what people choose, so we keep the raw answers, version the reason codes and check them against what the same subscribers did next.

Who owns it once it’s running?

You do. It’s built in your accounts, documented, with training for the people who run it. Ownership usually sits with your head of retention or lifecycle. We can keep running it with you, or step back.

How do we start?

Usually with a retention review: two to four weeks, a fixed fee credited to the build, and a plan that’s yours. If the job is already clear, we can scope the build directly.