The scenario
The brand has sold sunglasses, prescription frames and coloured contact lenses online for five years, with a WhatsApp list built mostly through checkout opt-in and a smaller number of orders placed directly by chat. The team had one habit for the whole list: a monthly message with a flat discount, sent to everyone regardless of what they had ever bought.
The three products rebuy on completely different clocks. Coloured lenses are a disposable product bought again every couple of months. Prescription frames renew on a much longer cycle tied to when the customer's own prescription changes. Sunglasses are close to a one-time purchase for most buyers, except as a gift bought again for someone else around Eid. None of that difference showed up in the messaging.
- Monthly revenue band
- pending client sign-off
- WhatsApp list size band
- pending client sign-off
- Prescription handling
- Customer's own optician prescription, lenses fitted by a partner optical lab
- Channel
- Official WhatsApp Business API, not a personal number
One monthly blast for a disposable product, a renewal product and a gift product
The monthly discount message went to the full list on the same day every month, whether the recipient had bought coloured lenses six weeks ago and was due to reorder, or bought a pair of sunglasses two years ago and had no reason to think about the brand again until a wedding came up. Repeat purchase across the list had been flat for over a year, and the opt-out rate on the broadcast had been climbing slowly the whole time.
Splitting the customer base by first product told a clearer story than the blended number ever could. Lens buyers who never received a reminder timed to their own reorder point simply drifted to whatever was easiest when their lenses ran out, often a different seller entirely. Frame buyers, whose actual moment of need was a year or two away, had been receiving a message every month that had nothing to do with them and every reason to eventually mute it.
The list was being managed as one audience because it had always been one list, not because the three products behaved the same way. A message cadence built around the calendar rather than the product's own reorder point was training the most loyal segment — lens buyers — to shop elsewhere, while wasting the same message on frame and sunglasses customers who had no reason to act on it yet.
What we did — the messages and flows
The losers are here on purpose. A test with only winners was never a test.
The lens reorder check-in, timed to the actual product
Winner"Your lenses usually last around two months. Ready for the next pair?"
- Format:
- WhatsApp template message, sent individually at each customer's own reorder point, Kuwaiti dialect
Built directly from the cohort read, which showed lens buyers reordering on a fairly consistent personal cycle that the brand had simply never messaged against. Timing the check-in to each customer's own likely reorder point, rather than a calendar date shared by the whole list, made this the single best-performing flow in the study and the clearest source of the repeat-rate movement.
The prescription check-in, timed to the renewal window
Winner"It has been a while — worth checking if your prescription has changed?"
- Format:
- WhatsApp template message sent around eighteen months after the last frame purchase, linking to the product page's prescription-upload step
Written to prompt the customer to check with her own optician, never to suggest her vision had changed or that the brand could judge that. It performed well precisely because it asked a genuine, low-pressure question rather than pushing a discount, and it moved frame repeat purchase from close to zero to a meaningful share of the line's revenue.
The Eid sunglasses gifting reminder, sent only to past sunglasses buyers
Winner"Eid is close. Someone on your list needs new sunglasses."
- Format:
- WhatsApp broadcast sent four weeks before Eid, segmented to customers who had bought sunglasses at any point, never sent to frame-only or lens-only customers
Reframes the sunglasses repeat purchase as a gifting occasion rather than a personal reorder, which matches how this product actually gets bought a second time. Segmenting it to sunglasses buyers only, rather than the whole list, is what kept it from becoming another version of the generic blast it was designed to replace.
The monthly flat-discount blast to the whole list
Lost"10% off everything this month."
- Format:
- WhatsApp broadcast sent to the entire list on the same date every month, unsegmented
Kept running during the first month of the engagement as the control the new flows were measured against. It carried the highest opt-out rate of anything sent and the lowest revenue per recipient, which is the clearest evidence that a shared cadence across three products with three different reorder clocks reads as noise rather than a reason to buy.
The 90-day lapsed lens buyer win-back
Neutral"It has been three months. Still wearing lenses?"
- Format:
- WhatsApp message sent once a lens buyer passes ninety days without reordering, offering a small specific reason to return rather than a discount
Response was modest and inconsistent, which reads as a genuinely mixed result rather than a clear win: some lapsed buyers had simply switched products or stopped wearing lenses altogether, and a single message cannot distinguish between someone who forgot and someone who has moved on. Worth another round with a shorter lapse window before drawing a conclusion.
What we did — the optimizations, in order
Read repeat rate by first product, not as one list
Pulled repeat purchase rate at thirty, sixty and ninety days, split by whether the customer's first purchase was sunglasses, frames or lenses.
Why: A blended repeat rate across three products with three completely different reorder clocks reads as flat even when one line is actually improving and another is quietly leaking. The split by first product is what revealed lenses as the gateway product worth building a flow around first.
Confirm the channel is the official WhatsApp Business API
Verified the existing WhatsApp presence ran on the official Business API rather than a personal number, and added a second opt-in prompt on the order-confirmation page for customers who had not opted in at checkout.
Why: A retention plan built on an unofficial number or tool risks the whole list to a ban with no warning, taking every flow down with it. Confirming the official API first is what let the rest of the work be built on something that would still be standing in a year.
Build one flow per product, not one flow for the store
Wrote and shipped three separate flows in Kuwaiti dialect — the lens reorder check-in, the frame prescription check-in and the sunglasses gifting reminder — each triggered by a different signal rather than a shared calendar date.
Why: The three products are rebought for three different reasons at three different intervals, so a single flow written for "the store" would always be right for one product and wrong for the other two. Building per product is what let the frame message ask about a prescription and the lens message ask about a reorder without either sounding out of place.
Set the cadence per product, not per calendar
Lens messages triggered around each customer's own reorder point, frame messages around eighteen months after purchase, and sunglasses messages tied to Eid and wedding season rather than a fixed monthly date for everyone.
Why: The old monthly cadence trained the most valuable segment — lens buyers who reorder every couple of months — to expect an irrelevant message most of the time, which is what teaches a list to stop reading. Matching cadence to each product's own rhythm is what turned the message from noise into something worth opening.
Match the offer to how each product is actually rebought
A plain reorder check-in for lenses with no discount attached, a check-in question for frames rather than any offer at all, and a gifting frame for sunglasses rather than a personal reorder message.
Why: Lens buyers did not need a discount to reorder something they were already going to buy again; what they needed was the reminder itself. Attaching a discount to every message would have trained the list to wait for one, which is a more expensive habit to break later than the one it replaced.
Build a specific win-back for lapsed lens buyers
A single message triggered ninety days after a lens buyer's expected reorder point passed without an order, offering a specific reason to return rather than a blanket discount sent to everyone regardless of what they had lapsed from.
Why: A generic lapsed-customer discount trains everyone to wait for the next one rather than reordering on time. Building the win-back specifically around the lens reorder cycle kept it targeted, even though the early result on this particular flow was modest enough to need another round.
Measure each flow on its own, and prune the blast
Tracked repeat rate, revenue per recipient and opt-out separately for each of the three flows against the old monthly blast, and stopped the blast once the flows showed a consistently better result on every measure.
Why: The blast had never been measured against anything, so it survived purely on habit. Measuring the flows against it directly, rather than assuming a segmented approach would obviously win, is what gave the decision to stop sending it a number behind it instead of a guess.
What changed
The headline figure sits in the table above, and the more useful read is which flow drove it: the lens reorder check-in carried most of the repeat-rate movement on its own, because it was solving the clearest problem — a genuinely repeat product with no reminder attached to its actual reorder point.
The frame prescription check-in moved a smaller number in absolute terms but from close to zero, which matters because that line had effectively no retention mechanism at all before this. Framing it as a question to ask her own optician rather than an offer is what made it land as helpful rather than as a sales message.
Opt-out fell once the blast was replaced, which is the clearest sign the old cadence had been actively costing the brand list members rather than simply failing to grow it. The lapsed-lens win-back remains the one open question in the results, and it is being treated as such rather than folded into the headline number.
What we would do next
Shorten the lapsed-lens win-back window from ninety days to closer to the customer's own expected reorder point plus two weeks, since a single fixed window for everyone repeats the same calendar-versus-product mismatch the original blast had.
Add a light referral prompt to the sunglasses gifting flow specifically, since a customer who bought sunglasses as an Eid gift for someone else is a plausible source of a second household on the list, and that flow has not been tested for anything beyond the original purchase yet.