How do you win back customers who haven't returned?

A customer who no longer comes back isn't necessarily lost. They may simply have changed their habits, spaced out their purchases, found another solution, or had no recent reason to return. The difficulty is therefore to tell a normal absence from a real change in behavior, and then to choose a suitable follow-up. An inactive customer is one whose absence has become unusually long compared with their expected purchase rhythm. And that length of time isn't the same for every business or every customer.

When does a customer become inactive?

There is no universal threshold. In a business where customers buy every week, 30 days without a purchase can already be a signal. In a store where purchases are naturally several months apart, the same duration would mean nothing. A fixed threshold such as:

“Inactive customer = no purchase in 90 days”

can be useful as a simple rule, but it must match the actual purchase cycle of the business. A more precise approach is to ask: has this customer been away much longer than usual? Take a customer who used to make about two purchases a month. If they haven't come back in three months, their behavior has clearly changed. Conversely, three months without a purchase can be perfectly normal for a customer who usually buys twice a year.

Which criteria help detect a customer drifting away?

The date of the last purchase matters, but it isn't always enough. Several pieces of information can be combined.

Recency

How long ago was the last purchase? This is the first indicator for spotting possible inactivity.

Historical frequency

How often did the customer buy before? It puts their current absence in the context of their usual behavior.

Regularity

Two customers who made twelve purchases during the year can behave very differently. The first may buy about once a month. The second may have made all twelve purchases over three months and then never come back. The annual frequency is the same, but the inactivity signal is not.

Amount spent

The revenue generated by the customer measures their past commercial weight over a given period. It should not, however, be confused with their overall “value” or profitability.

Products or categories purchased

Purchase history can help you understand the customer's profile and, where relevant, prepare a more relevant message.

Not all inactive customers need the same follow-up

Sending the same message to every customer who hasn't bought anything in six months is easy. It isn't necessarily relevant. Consider three profiles:

Customer A

A single purchase eight months ago.

Customer B

One purchase a month for two years, then nothing for three months.

Customer C

Several large purchases the previous year, but a naturally irregular frequency.

They are all “inactive” according to some thresholds. Yet their situations aren't comparable. Customer B shows a real change in behavior. Customer A may never have been a loyal customer. For customer C, you need more context before concluding they are drifting away. The quality of the follow-up therefore depends first on the quality of the segment — the same criteria used in customer segmentation to tell apart profiles that are still active.

Which customers should you follow up with first?

The answer depends on the business's goal. You can look for, for example:

• former regular customers who no longer come back;
• customers who generated significant revenue and whose activity has stopped;
• usual buyers of a specific category;
• customers whose absence clearly exceeds their usual rhythm.

It is often more useful to work on a well-defined segment than to follow up with the whole inactive base. Customer segmentation lets you build these groups based on measurable criteria.

A five-step follow-up method

1. IdentifySpot the customers whose absence is becoming significant.
2. SegmentTell profiles apart by their past behavior, frequency, recency, purchases or the amount generated.
3. Choose a reason to come backBefore writing the message, decide what the business really has to offer: something new, relevant information, a perk, a commercial offer or a simple reminder.
4. Follow upThe message and channel are adapted to the chosen segment.
5. MeasureAfter the campaign, check how many customers actually came back and under what conditions.

This last step is essential to improve future follow-ups.

Do you have to offer a discount?

No. A discount can be effective in some cases, but it shouldn't become the automatic reflex. It reduces your margin and can get some customers used to waiting for a promotion before coming back. Other approaches are possible:

The choice depends on the segment, the business and the presumed reason for the drop-off.

Personalize without over-interpreting the data

Purchase history lets you tailor a follow-up, but avoid drawing excessive conclusions from it. A customer who bought a product category three times doesn't necessarily want to receive only offers about that category. The data shows what happened, not with certainty what the customer will want tomorrow. Good personalization therefore uses history as a signal, not as an absolute truth about the customer's preferences.

What message should you send to an inactive customer?

The message should above all give a clear reason to come back. A simple structure can work:

1. recall the relationship with the business;
2. state the reason for the message;
3. suggest a simple action.

For example:

“It's been a while since we last saw you. We've just received our new [category] collection. Come and see it in store.”

Or, when an offer is truly justified:

“You used to buy [category] regularly. We're keeping a perk for you on this selection until [date].”

The message must remain consistent with what is actually known about the customer.

Choose the right timing

Following up too early can be pointless. Following up too late reduces the chances of reviving a behavior that stopped long ago. The right moment therefore depends on the purchase cycle. For a customer who normally comes in every two weeks, waiting six months probably makes no sense. For a purchase made once or twice a year, a follow-up after a few weeks would be premature. History lets you compare the time elapsed since the last purchase with the usual rhythm.

Beware of commercial pressure

Reactivating a customer doesn't mean multiplying contacts. Someone who doesn't respond to several campaigns shouldn't necessarily receive more and more messages. The frequency of communications must stay under control and respect the choices and consents that apply to the channel used. A relevant follow-up is generally better than a series of generic messages.

How do you measure a reactivation campaign?

Open rate or click rate can tell you something about the message. But the end goal is generally the customer's return. It is therefore useful to track, depending on the available data:

• the number of customers contacted;
• the number of customers who came back;
• the time before the new purchase;
• the revenue generated after the follow-up;
• the basket of the returning customer;
• possibly the margin generated.

You can then calculate a reactivation rate:

Reactivation rate = customers who came back ÷ customers contacted × 100

The period used to consider that a customer has “come back” must be defined in advance. Without this observation window, the result becomes hard to interpret.

Test rather than assume

Two messages can seem equally relevant without producing the same result. When the number of customers allows it, it can be worthwhile to compare different approaches:

• with or without a perk;
• something new or a reminder;
• different send times;
• different segments.

Over time, this lets you understand which follow-ups really work for the business. The goal isn't to find a universal formula, but to improve your strategy based on observed results.

Automate detection without blindly automating communication

Searching for customers who are starting to drift away is repetitive work. It can therefore be automated. See Automating your store's tasks for details on how routines work. For example, a system can regularly check which customers exceed an inactivity threshold or move away from their usual frequency. However, automatically detecting an inactive customer doesn't mean a message must necessarily go out automatically. It may be preferable to keep an approval step before certain communications, especially when the segment or the offer needs a check. This distinction lets you automate monitoring without losing control of the commercial action.

How can AI help?

When it has the necessary data, an AI can make it easier to search for and analyze the customers concerned. For example:

“Which customers who used to buy at least once a month haven't bought anything in three months?”

Then:

“Among them, which ones regularly bought category X?”

And finally:

“Prepare a reactivation message for this segment.”

The benefit is being able to move step by step from detection to targeting and then to preparing the action, without reducing all inactive customers to a single group.

How can Gillia help?

Gillia lets you query the available customer data to identify drop-off behaviors. The merchant can, for example, ask:

“Which regular customers haven't bought anything in 90 days?”

Then refine:

“Keep those who made at least six purchases over the previous twelve months.”

Once the segment is checked, Gillia can help prepare the matching message from the customer loyalty capability. When a search needs to be repeated regularly, it can also become a routine that surfaces new customers matching the criteria. Detection can thus be automated, while the list, the message and the action stay under your control before sending — the principle of a true AI agent for retail.

Reactivate rather than simply follow up

A good reactivation strategy isn't about sending more messages. It's about identifying:

• who really shows a change in behavior;
• when that change becomes significant;
• what relevant reason can be given for them to come back;
• and whether the follow-up actually brings them back.

This logic is what turns a base of inactive customers into a genuine loyalty approach. More loyalty examples can be found in our catalog of Gillia use cases: identify and follow up with inactive customers or get alerted when a regular customer drifts away.