The Markin ROI Report for Enterprise Growth TeamsRead now
MARKIN
Field notes
Playbooks9 min read

Customer winback strategies that pay for themselves

Most win-back programmes mail the whole churned base and count natural returners as wins. How to segment, price the offer and prove causation.

Elisa Fontaine
  • #Churn
  • #Retention
  • #Playbooks
Customer winback strategies that pay for themselves

Customer winback strategies are the actions a subscription business uses to reacquire customers who already cancelled: win-back campaigns, come-back offers, reactivation journeys. The ones that work treat win-back as an uplift problem: who is reachable, which offer is margin-positive for this specific person, and a holdout that proves the campaign did something.

Most win-back programmes fail on one of those three. They mail everyone who ever cancelled, they lead with the deepest discount in the building, and they count every resubscription as a win, including the customers who were coming back on their own.

Why win-back is not one population

A churned base is at least three different groups wearing the same label:

  1. 1Recent churners with intact habits. Cancelled in the last 30 to 90 days, were active right up to the exit, and left over a specific event: a price rise, a content gap, a failed payment nobody fixed. This is the only segment where win-back economics are reliably positive.
  2. 2Seasonal and cyclical leavers. They leave and return on their own schedule: sports packages, travel passes, tax software. Contacting them is mostly wasted spend, because the majority return anyway. The right move is to identify them and spend nothing.
  3. 3The long gone. Churned a year or more ago, contact details decaying, habit replaced by a competitor. In our datasets, response rates here run at a tenth of the recent-churner segment, and the few who return churn again at twice the base rate.

Reachability is the first filter, not the message

Email addresses decay. Roughly a third of churned-customer contact data becomes undeliverable or unengaged within 18 months, faster in prepaid-heavy categories. Before any creative or offer work, score each churned customer on reachability: valid and consented channel, recent engagement with transactional mail, and an active app install or login. The unreachable segment should get paid suppression, not campaigns.

The margin-positive offer

The right win-back offer is the cheapest action that changes the decision for this customer, and it is almost never the sitewide 50 percent off. Rank the candidates the same way you would rank any retention action, by uplift-weighted margin:

  1. 1Fix the exit reason. If they left over a price rise, the offer is the old price or a right-sized plan. If they left over a product gap that is now closed, the offer is the news. Reason-coded exits, captured in a well-built cancellation flow, are what make this possible.
  2. 2Time-limited, single-use incentives. A returning customer who is trained to expect a win-back discount will churn on schedule to collect the next one. Cap the incentive, cap the frequency, and never run an always-on win-back offer the base can learn.
  3. 3Nothing. For seasonal leavers and for customers whose predicted second-tenure value is below the cost of contact, the profit-maximising win-back action is no action.

Prove it with a holdout

Natural return rates are not zero. Between 5 and 15 percent of churned subscribers in high-frequency categories come back within a year with no contact at all. If your win-back reporting counts them, your programme is taking credit for gravity. Reserve a random slice of every win-back audience as an untreated control, and report incremental resubscriptions and incremental second-tenure margin, not raw conversion.

Second-tenure quality matters as much as the return itself. Measure won-back customers on 90-day retention and on ARPU relative to their first tenure. A win-back that returns customers at half their previous value and double the churn rate is buying revenue, not earning it. The same measurement discipline applies to pricing and plan changes on the returning cohort, which is where win-back connects to ARPU growth rather than sitting in a separate reporting silo.

A 90-day build order

  1. 1Segment the churned base by recency and reason. Join the cancellation reason from the exit flow to billing and engagement history. Drop everything older than your reachability horizon.
  2. 2Score uplift, not propensity. A model that ranks who is likely to return will target the seasonal leavers who return anyway. A model that ranks whose return probability changes when contacted targets the persuadable middle.
  3. 3Launch with the holdout built in. Ten percent untreated, read at 30, 60 and 90 days, second-tenure margin as the acceptance metric. Kill any segment where incremental margin is negative, including segments that convert well.

Markin ranks churned customers on uplift-weighted margin and reserves a holdout on every win-back audience, so the programme reports returns it actually caused. See Retention decisioning or the streaming view, where win-back volume is highest.

Frequently asked

Questions readers ask about this.

What is a customer win-back strategy?
A win-back strategy is the set of decisions used to reacquire customers who already cancelled: which churned customers to contact, through which channel, with which offer, and how to prove the contact caused the return. It is distinct from churn prevention, which acts before the cancellation, and from deflection, which acts inside the cancellation flow.
How long after churn should you run win-back?
The profitable window is usually the first 90 days after cancellation, while habits and contact data are still intact. Response rates on customers churned more than a year ago run around a tenth of the recent-churner segment, and those who do return churn again at roughly twice the base rate.
What is a good win-back offer?
The cheapest action that changes the decision for that specific customer. If they left over a price rise, a right-sized plan beats a discount. If they left over a product gap that has since closed, the news is the offer. Deep sitewide discounts train the base to churn on schedule to collect them.
How do you measure win-back correctly?
Reserve a random untreated control from every win-back audience and report incremental resubscriptions and incremental second-tenure margin, not raw conversion. Natural return rates of 5 to 15 percent mean uncontrolled reporting takes credit for customers who were coming back anyway.
Are won-back customers as valuable as new ones?
Not automatically. Track second-tenure ARPU against their first tenure and their 90-day retention. A win-back that returns customers at half their previous value and double the churn rate is buying revenue rather than earning it.

Related resources

Bring this into your workspace.

Solutions

Industries

See it in the product

This runs in Markin today.

The same loops this note describes run 24/7 against your customer base. Watch the workspace decide, experiment and execute 1:1.

Explore demoOr book a demo with the team

Vocabulary

The terms behind this page.

Definitions an answer engine can quote, each one a page of its own.

Full glossary

Compare

How this compares to what you already run.

Neutral reads on the categories buyers evaluate against a decision layer.

All comparisons

Keep reading.

ResearchJul 27, 2026

Churn prevention economics in streaming and telecom

PlaybooksAug 07, 2026

Customer churn prediction: models, features and incrementality

GuidesAug 21, 2026

Voluntary vs involuntary churn: two problems, two fixes