How to win back lapsed customers with a three-email flow
By Toby Dames. Published 5 November 2026.
Step 1: define lapsed from your own data
Find the median days between a first and second order, and between later orders. A customer is lapsed when they pass that point by a margin, for example one and a half times the median. For a brand where customers reorder every 60 days, someone at 90 days is overdue. For a furniture brand it may be a year or more.
Avoid a fixed rule such as "180 days" unless your data says it fits.
Step 2: segment before you send
- Best customers who have gone quiet. Highest value, so they get the most personal message.
- Single-order customers. They need a reason to try again.
- Discount-only buyers. Be careful with offers here, or you teach them to wait.
Step 3: a three-email sequence
- Day 0. A check-in. Short, human, no offer. Something that sounds like a person wrote it.
- Day 3 to 4. What is new. New products, restocks and reviews since they last bought.
- Day 7. A reason to return. An offer or a gift, sized against margin. For your best customers, use access or a gift rather than a percentage off.
Stop the sequence the moment someone orders. Suppress people who have unsubscribed, and take care with SMS consent under the Spam Act.
Step 4: check it paid for itself
Hold back a random 10% of eligible customers who receive nothing. Compare orders from both groups after 30 days. The difference is what the flow produced. Subtract the offer cost and your margin, not just the revenue.
Without a comparison group, customers who would have returned anyway get counted as wins.
Questions
How many emails should a win-back flow have?
Three is a good start. Test adding a fourth only if the third still earns more than it costs.
Should every win-back email include a discount?
No. Lead with the check-in and what is new, and keep the offer for later in the sequence. Offers can train customers to wait, and they cost margin on people who would have returned anyway.
What if my list is small?
The hold-out comparison gets noisy with small lists. Compare against your recorded starting figures instead, and treat the result as indicative.
Want this worked out on your own data?
The free Profit Leak Detector takes a couple of minutes. Or book 30 minutes and we will look at your numbers together.