When has a customer stopped buying?

Most win-back rules start with a number of days. This guide explains why one number doesn’t fit a store, how Shopify and Klaviyo frame the question, and how to judge each customer against their own pattern. A lapsed customer is a customer who has gone quiet for longer than usual, so the useful measure is “longer than usual for them”.

Why one number fits no store

A fixed cut-off such as 90 or 180 days treats every product the same. This is our reasoning, not a published finding. A customer who buys coffee every week has gone quiet after a month. A customer who buys a mattress every few years hasn’t gone quiet after a year. One number flags the first too late and the second too early.

A quiet customer might be between orders or gone for good, and nothing on their record says which. A longer silence than usual is the evidence you have, so measure it against the customer’s own pattern.

How Shopify and Klaviyo frame it

Neither tool uses a fixed day count as its main idea.

• Shopify RFM. Shopify scores each customer from 1 to 5 on recency, frequency and monetary value. It says “RFM scores are based only on data from your store, and not on industry standards or third-party data”, and that a 5 puts the customer “in the top 20% of the dimension for your store”. Recency is relative to your other customers. Two of its groups describe customers who have gone quiet: Dormant (“Customers without recent purchases, with infrequent orders, and with low spend”) and At risk (“Customers without recent purchases, but with a strong history of orders and spend”). Source: Customers reports.
• Klaviyo. Klaviyo defines average time between orders as “The average number of days between each of a customer’s orders”, and says its risk-of-leaving prediction “is based on their number and frequency of orders” (Predictive analytics). Each customer is compared with their own pattern.

Work out a customer’s usual gap

Take a customer with several orders and measure the days between each pair. The usual gap is the average of those. Here is an invented example:

The multiple is your choice. A small multiple flags people early and brings more false alarms. A large one flags them late and misses people you could still reach. Test it on your own orders (see the end of this guide) before you send anything.

Add a floor for one-time buyers

A customer with one order has no gap to measure. Give them a floor: a fixed number of days, set from how long your product normally lasts, before they count as stopped. This is a judgement for your store. A pet food store and a furniture store will choose very different floors.

Check for a second profile first

A shopper who starts using a new email gets a new profile. Their old profile goes quiet, and a rule that reads profiles will list them as stopped. In MergeGuard’s demo store: 905 customers look like they stopped ordering → 842 really did; 63 still buy on another profile Demo store (Northsail Supply). Those 63 are the false alarms. A discount sent to them goes to people who never left.

Checking for a second profile means looking for a recent order, on any profile, from the same person. A shared phone or an email that differs by a typo are strong signs. A shared address alone is a reason to look, and not proof. Counting people for this check doesn’t require a merge.

Use the list

Once you have a list of customers who really stopped, send it to your own email tool. MergeGuard never sends messages. Decide the offer for each group separately, since a customer who stopped after one order and one who stopped after twenty have different reasons.

Test it on your own orders

A simple test shows whether a gap-based rule beats a fixed cutoff for your store.

1. Build two lists from the same date: customers over a fixed 90 days, and customers past your multiple of their usual gap.
2. Wait 30 days without contacting either list.
3. Count how many people on each list ordered during those 30 days.

The better rule flags fewer customers who then order anyway. You set the numbers, and the result applies to your products and your customers.