Recall: 2,173 Customers Are Due, and Exactly Two Have Ever Been Reminded

There is a list in your shop of everyone who has ever bought a pair of spectacles from you. In most shops it is a bill book; in ours it is a database with 4,412 people in it, every one with a phone number.

I ran one query against it that I am not going to forget. Of those customers, 2,173 have a recall date already recorded — a date somebody sat down and entered, meaning contact this person around here. The number who have ever actually been sent a recall reminder is two.

Not two percent. Two people.

Only one customer in five ever comes back

Start with the honest baseline, because recall is usually sold with a fantasy attached.

Of 2,091 customers who have bought at least once, 456 — 21.8% — have bought more than once. 159 have bought three times or more. Four out of five people who walked in, chose frames, waited for the lab and collected their spectacles have never been back.

And yet returning customers are not a rounding error in the takings. Of the 1,059 orders placed since January 2025, 339 — 32% — came from someone who had bought before.

That is the whole argument for recall in two numbers. A fifth of the customers produce a third of the orders. Every percentage point you add to that fifth is worth more than the advertising it would take to replace it, because these people have already chosen you once, and you already have their number.

Half the list has already gone quiet

Now the part nobody prints on a marketing slide:

• 48.4% of customers have not bought anything in two years.
• 64.7% have not bought in eighteen months.

Two-thirds of the list is cold. That is not a reason to leave it alone — it is the reason the list is worth working at all. A shop that recalls at the right moment does not have a two-thirds cold list, because people are being caught before they drift.

It is also a warning about how you re-enter the conversation. Somebody last seen in 2023 should not receive the same message as somebody who collected last month, and if the first thing you say after two years of silence is an offer, you have told them exactly what the relationship is worth.

The recall window is two windows, not one

Every optical software demo in India shows a twelve-month recall. So I measured the gaps between consecutive purchases by the same customer — 698 of them.

The median gap is 165 days — about five and a half months. The upper quartile is 367 days.

Look at the shape rather than the average, because the average is describing two different things stuck together:

• A short spike: 22.2% of return visits happen inside a month. That is not a new prescription. That is the second pair, the spare, the sunglasses they decided on afterwards, the spouse they brought in, the frame that did not suit at home.
• A long hump between six months and two years, 41.3% of return visits, which is the prescription-driven return everyone actually means by "recall".

A single twelve-month reminder is aimed at the gap between the two. It arrives too late for the 36% who would have come back within three months, and too early for the fifth who return after a year and a half. The same shop needs two different conversations — one in the weeks after collection, one at the eye-test anniversary — and they are not the same message, the same list, or the same reason to write.

Having the number is not the same as being allowed to use it

Here is where a lot of Indian optical shops are about to walk into a wall.

Every one of those 4,412 customers has a phone number. On the WhatsApp side, the service flag is on for all of them — which is what you would expect, since it was defaulted on at migration — but marketing consent is recorded for exactly 0%.

Those are two different permissions, and the difference is not academic:

• A service or utility message relates to something the customer did: their job is ready, their order is billed, their appointment is confirmed.
• A marketing message is anything that reminds, tempts or offers — which is what a recall usually is.

You need recorded consent for the second kind, the customer must be able to stop it, and from 1 October 2026 the economics change too: WhatsApp service messages stop being free, with 1,000 free per business number per month and a per-message rate after that. The full set of rules, and what gets a number banned, is in the WhatsApp article.

So the first job of a recall programme is not the message. It is a consent tick at the counter, dated, with a record of where it came from — and after that, a recall list of people who agreed to hear from you is worth more than a list of everyone.

What a recall actually looks like

Not a blast. A blast is what shops do when they have a list and no system, and it is how numbers get reported and blocked.

What works is small, dated and reason-specific:

1. Two weeks after collection. Is everything comfortable? No offer. This one catches the adjustment nobody comes back for and it is the message that earns the right to send the rest.
2. At the prescription anniversary — the date your own test recorded, not a generic twelve months from the bill. If you kept the test date (most shops do not), this message writes itself.
3. Contact lens re-supply, on the wearer's actual pack cycle — monthly, quarterly — which is the most predictable recall in the shop and the one most often left to the customer to remember.
4. Warranty and coating, where you have promised something with a date on it.
5. The quiet ones, deliberately last: people eighteen months out, once, with a reason to come in that is not a discount — a free check of the existing pair, for instance.

Three rules that decide whether any of this works: stop the sequence the moment the person walks in; keep quiet hours; and count replies, not sends. A recall programme is judged by how many people walked through the door, and that number belongs next to the bill, not in a messaging dashboard.

The arithmetic, stated as arithmetic

I am not going to claim a return figure, because I do not have one to claim — we have sent two reminders. But the arithmetic that makes recall worth the effort is simple and you can run it on your own numbers tonight.

This shop takes about 800 orders a year, 32% of them from returning customers. If a working recall moves the repeat rate from 21.8% to 26% — four points, on a list where two-thirds are already cold and have never been contacted once — that is roughly 80 more orders a year from people who already own your spectacles. No advertising, no discount, no new footfall.

Whether it moves four points or one, I will report honestly when the shop has actually run it for a year. What is not in doubt is the starting position: 2,173 recall dates sitting in a database, and two reminders ever sent. The list was never the problem.

Counts from one Indian optical business with two branches: 4,412 customer records and 2,883 orders from 2,091 customers, September 2022 to 2026, as at 18 September 2026. Return-gap figures use 698 consecutive-purchase intervals. Consent figures are the recorded flags in that database. No customer details, order values or branch identities are published.