What changed · Retail
Nobody notices when a regular stops coming
The most valuable customer you will lose this year will leave without anyone noticing, because nothing in a store is watching for absence.
20 August 2026 · 5 minute read
Every business loses customers gradually rather than suddenly, and the ones worth keeping give plenty of warning. An e-commerce brand sees that warning automatically: the last order date sits in a column and a query finds everyone who has drifted. A brand selling through stores has no such column, so a customer who came every month for three years and has not been in since March is indistinguishable from one who came in last week. The absence is the signal, and absence is the one thing a system built around transactions never records.
Why absence is harder than it sounds
Systems record events. A sale happens and a row appears. Nothing happens when a sale does not happen, so there is no row, and no report built on rows will ever show it to you.
This is why sales dashboards are reassuring right up until they are not. Total revenue holds steady while the composition underneath changes: new customers replacing quietly departed regulars, at a higher acquisition cost, with lower lifetime value. The chart looks flat. The business is deteriorating.
What lapse actually looks like by category
The window that counts as "gone quiet" is entirely category-specific, and getting it wrong in either direction wastes money.
| Purchase pattern | Normal gap | Worth a nudge at |
|---|---|---|
| Groceries and daily use | Days to two weeks | 3 to 4 weeks |
| Personal care and cosmetics | 4 to 8 weeks | 10 to 12 weeks |
| Apparel | Seasonal, 3 to 4 months | 6 months |
| Footwear | 6 to 12 months | 14 months |
| Consumer durables | Years | At the service or replacement point |
| Restaurants and cafes | Days to weeks | 4 to 6 weeks |
Work yours out from your own data rather than from this table. Take everyone who bought twice, measure the gap between their purchases, and the point where most of the distribution sits is your normal. Roughly double it is where concern starts.
The counter-intuitive part
The right people to contact are not the ones who have been gone longest. Someone absent three years has moved on. The valuable group is the one just past normal: still in the habit, not yet replaced you, and cheap to bring back. Most win-back campaigns target the wrong end of the list because it is the more dramatic-looking number.
What to send, and what not to
The reflex is a discount, and it is usually the wrong first move. A discount to someone who was about to return anyway is a straight loss, and it teaches everyone else to wait.
- What tends to work: something new since they last came, a genuinely useful reminder tied to what they bought, or a reason connected to a moment they already care about.
- What tends not to: a generic offer, a loyalty points balance nobody understands, or anything that opens with "we miss you".
- The channel that matters in India: WhatsApp, because it reaches people email does not, and because a message there feels like a shop remembering you rather than a database processing you.
Why this rarely gets built
Because it requires knowing when each person last bought, which requires the one-file exercise, which requires three departments to hand over exports. Compared with running another festival campaign, it is unglamorous and politically annoying.
It is also worth more. A brand that contacts customers at the right point in their own cycle is doing something none of its competitors are doing, using data it already had, at close to zero media cost.
Revenue can hold flat for four straight quarters while a business quietly replaces its best customers with expensive new ones. Nothing on a sales dashboard shows this, because absence never produces a row.
If you are an agency
This is a strong retained proposal for a retail client because it recurs monthly and improves as the list grows. Define the lapse window from the client's own repeat-gap data rather than a category rule of thumb, and report on the group just past normal rather than on the long-lapsed, which is where the recoverable value sits. It also gives you a WhatsApp workstream that is genuinely useful rather than promotional.
How we do this ourselves
The grouping identifies who has gone past their normal gap and puts them in their own group, and WhatsApp broadcasts with scheduled sends are built in, so the message goes out on the channel that actually reaches people in India.
How lapse groups are built