Churned-Account Win-Back Playbook 2026

Use the benchmark from the weekly check, or build one: 20 or so customers in the same segment, 12+ months in, renewed at the same or higher revenue, with a core measure your CS lead approves (default: share of seats active in three of the last four weeks, and the features they use weekly). With fewer than 20, label it low confidence.

For each former customer, pull four fields from the CRM, billing and PostHog, Amplitude or Mixpanel:

• Reason they left: the exit reason logged at cancellation. If you don't log one today, start now with six to eight codes, and backfill the last year from CRM notes.
• Usage at exit vs benchmark: their 90 days before cancellation, measured the same way.
• ARR when they left.
• Who made the call: the person who signed the cancellation.

Say an account paid $14,000 and left in March because it couldn't schedule exports. In its last 90 days, 11 of 14 seats were active (79%), where the benchmark is 70%. In August, scheduled exports ship. The $14,000 is a comparison, not a forecast. (The numbers are illustrative.)

Run it as a saved query. Under about a hundred former customers, do it by hand.

• Fit at exit (defaults to tune): Close – at or above the benchmark in their last 90 days. Partway – reached it once, then slid. Never – never reached it.
• Fixed or repriced since? List each release from your Beamer changelog and each pricing change against the exit reasons it answers, in the sheet. A match names the fix.
• Months since they left, from the cancellation date.
• Decision-maker still there? Check email bounces and their public profile. If they've gone, address the note to the most active user at exit.

If they never reached the benchmark, check why. Revisit them only when the change addresses the blocker that stopped them getting value.

Leave it alone if any of these hold (defaults to tune): fit was Never and nothing since addresses what stopped them; the exit reason isn't a product gap or price (a relationship, closed or acquired, other); the reason still stands; they left under 90 days ago; they left for a competitor under 12 months ago; they had a win-back note in the last six months; they unsubscribed or asked not to hear from you (check your email tool's opt-out list, not only the CRM). Accounts that pass pick a move.

Set Recommended move to No move, write the reason, and set Review date to when it could change: the next quarter, the competitor's renewal, or "when [reason] ships". See the leave-alone play. A note before the fix exists reads as denial; the same note after it ships reads as follow-through.

• A fix: the reason was a product gap and the changelog shows the fix.
• A new plan: the reason was price or packaging, and a new tier, plan or usage-based price answers it.

If both apply, send the fix note and add the plan in one line.

Four short sentences to the person who made the call: why they left, what shipped, what it means for the feature their team used most, and a small ask. No discount in the first note: lead with the fix, and keep price for a reply that raises it.

The same shape, about price: what they used, the plan that now covers it, and what it costs.

The AE reads the account's history, then sends the note as drafted, changes it, or passes. They answer any reply within one working day (default): the question is no longer what the product does, but whether it's safe to come back. Before sending, record the success measure and the review date: a reply is progress, a signed return is the result. If your team chooses to automate the fix note for smaller accounts, send it from Customer.io with replies routed to the AE and the send logged on the record. Before each win-back, check the recipient's location, the permission or exception that covers this message, and all recorded opt-outs. Don't assume a past purchase gives ongoing permission. Identify your business, include your postal address and give an easy way to unsubscribe.

On the review date, record the result, the win-back ARR, and what else played a part; the eight-week review has the mechanics. If nothing came back, change the note, stop, or give it time. Report resurrection revenue on its own line; folded into new business, the play's result disappears, and it's hard to fund next year. Each quarter, tell product which exit reasons came back, and for which fixes.