How the offer is decided
The default win-back offer is a percentage off, it is chosen because it is easy to configure, and it is usually both unnecessary and quietly expensive. The ladder starts somewhere much cheaper.
Key takeaways
- Rung one is a reminder with a question, and it converts more than people expect.
- Rung two is something useful that costs nothing: a guide, a reorder link, a slot.
- Rung three is a non-price sweetener: free delivery, a sample, priority.
- Rung four is a discount, and it is the last one for good reasons.
- Every reply goes to a person. A reply is worth more than the order.
The ladder
| Rung | What it is | Costs | Use when |
|---|---|---|---|
| 1. Reminder | “It has been a while — running low?” | Nothing | Consumables, habitual purchases |
| 2. Useful thing | A reorder link, a slot, a how-to | Nothing | Anything with friction to reordering |
| 3. Non-price | Free delivery, a sample, priority booking | A little | Where price was probably not the reason |
| 4. Discount | A percentage or an amount off | Margin, and precedent | Where you know price was the reason |
The last column is the one to read carefully. A discount is right when you have evidence that price was why they left — they said so, or a competitor undercut you visibly — and it is a guess in every other case. Guessing costs margin on everybody who would have come back for a reminder.
Rung one converts more than expected
A meaningful share of lapsed consumable customers simply forgot, changed a routine, or ran out at an inconvenient moment and bought something else once. For those people the reminder is the entire intervention and the discount is pure giveaway.
The message that works is a question rather than a pitch: “running low, or did you find something better?” Both answers are useful, the second one is worth more than the order, and asking it is what distinguishes this from a promotional email.
What a discount actually costs
The middle bar is what most win-back campaigns do, and on a recovery-rate dashboard it is the winner. The margin given away does not appear on that dashboard, which is how the practice survives.
The third approach — send the reminder, and offer a discount only to people who reply saying price was the problem — recovers nearly as much and keeps almost all the margin, at the cost of needing a person to handle replies.
The precedent cost
There is a second cost that is harder to measure and worth naming: a customer who receives fifteen per cent off for lapsing has learned that lapsing is worth fifteen per cent. Businesses that run a discount-led win-back for a couple of years reliably notice their best customers developing a suspicious rhythm.
That is not an argument against ever discounting. It is an argument for the discount being a response to a stated reason rather than an automatic consequence of not buying, which is a distinction the customer can feel.
Replies go to people
- App integration
- Machine learning
- Security & identity
- Analytics
- People
This is also why the message is signed by a person with a working reply-to. A no-reply address on a win-back converts the one honest channel into a broadcast, and the information that would have told you why fourteen per cent of your customers left goes into a mailbox nobody reads.
Next: counting what actually happened.
All posts