Who should not get a win-back discount

A win-back discount is a bet that a customer needs a push. It loses money when it reaches someone who was going to order anyway. The easiest way to make that mistake is to trust a list of people whose last order is old, when some of them are ordering under another profile. This guide shows who to take off the list before the offer goes out.

Why a list of people who stopped ordering overstates

Shopify makes a profile from the details a shopper types. A returning shopper with a new email starts a new profile, and the old one stops ordering. A report that counts by profile lists the old profile as a customer who left.

In MergeGuard’s demo store: 905 look like they stopped ordering → 842 really stopped ordering; 63 still buying under another profile Demo store (Northsail Supply) Some of the people on that list were still buying and would have been offered a discount they did not need. No external study measures this for other stores, and the demo is one store. Your numbers will differ, and counting people is the way to find them.

There is no universal number of days after which a customer has stopped. Klaviyo, for one, bases its risk prediction on “their number and frequency of orders” and reports each customer’s average time between orders (Klaviyo Help Center). Our reasoning is that each customer should be judged against their own usual gap. A weekly coffee buyer who has been quiet for two months differs from a mattress buyer who has been quiet for two years.

Who to leave out

Take these customers off the list before you send:

• Anyone with a recent order on any profile. The most recent order across all of a person’s profiles decides whether they have stopped.
• People you may not email. A list can include customers who never subscribed, or who unsubscribed after their last order.

Relatives and people who live together are separate customers, so judge each on their own orders. Do not pool their histories, and do not take one off the list only because they share an address with a customer who still buys. In MergeGuard you can mark relatives as family members.

Do not merge profiles just to clean a list. A Shopify merge is permanent, and a close call is safer left alone. Treating two profiles as one person for a count needs no merge.

Check subscription status before you use the list

Shopify’s customer segments page shows a subscription_status column in its default query, and its example prompt asks for “customers who last ordered more than 90 days ago and are subscribed to marketing emails.” The page says nothing about discounts or marketing exclusions, so it is a way to see the status, and the decision to email stays with you.

Whether you may email a customer is a consent question for you and your email tool.

Hold back a test group

You cannot tell from the list whether the offer worked unless some similar customers did not get it. Our suggestion, which the merchant can adjust:

1. Send the offer to half of the cleaned list, chosen at random.
2. Keep the other half on your normal emails.
3. After a period you set, compare margin per customer in the two halves, not only order counts.

Success is a higher margin per customer in the half that got the offer. No difference means no margin improvement was observed. People may still have behaved differently, because extra orders can be canceled out by the cost of the discount. Compare order counts and timing in the two halves as well. You choose the period and the size of difference that matters to you. We found no external study on the cost of discounting customers who were already active, so your own test is the evidence.

Where to send

Your own email tool sends the offer, so build the send there from the cleaned list. Help shows how to build and export a win-back list in MergeGuard.