Win-back
Some of the people who left will come back. Most will not.
Win-back is the easiest campaign to run badly: send everyone a discount, count the returns, ignore the ones who would have come back on their own. Tenure treats it like every other decision here — targeted, capped and measured against a control.
Left for a reason that has since changed, on a plan that still fits, inside the window where returns happen.
Worth asking, deliberately not asked. Some of them return anyway, and that is the number a campaign has to beat.
Chronic discount-seekers, people who churned in their first month, anyone already asked this quarter.
A hundred people who cancelled. An illustration of the shape, not a customer cohort.
Who is actually worth asking
A former subscriber is not a lead, and treating them like one is how you get unsubscribes.
- Segments built from why they left, how long they stayed and what they were paying
- Excludes the people your own data says will not return
- Timing rules, so the ask lands when it might land well
- Frequency caps, so nobody is asked twice in a quarter
Measured like everything else
Including the part most tools leave out: the people who would have returned anyway.
- A control group on win-back campaigns, not just on the cancel flow
- Return rate reported net of the control group
- Offer cost tracked against the revenue the return actually produced
- Whether returned subscribers stay, or churn again within a cycle
The rest of it
Everything reports to the same place.
Setup from your app
AI reads your site and billing, then writes the flow.
Cancel widget
The save flow that runs where people actually cancel.
AI agent access
Give a support agent the offer, and the limits it may not cross.
Offer reporting
Which discounts are worth what they cost you.
Payment recovery
Failed cards, and the subscriptions they quietly end.
Ask the right ones, and only once.
Win-back runs with its own control group, so a return is not counted twice.