The scenario
The brand sells silver and demi-fine necklaces, earrings and bracelets, direct to consumer, discovered mostly through Instagram and Snapchat and sold on a Shopify store with KNET, Tabby and cash on delivery at checkout. Five years in, one founder, no dedicated retention role, and a customer list of several thousand buyers that had never been messaged with anything beyond an order confirmation and a shipping update.
The revenue pattern was unmistakable once anyone looked at it by cohort: a sharp spike around Eid, a second around Mother's Day on the twenty-first of March, and almost total silence between them. The founder had assumed this was simply how jewellery sells in Kuwait. It was partly true and partly a self-fulfilling prophecy, because nothing was ever sent to a past buyer outside those two windows to give her a reason to come back.
- Monthly revenue band
- pending client sign-off
- Average order value
- pending client sign-off
- Fulfilment
- Own stock, gift-box packaging, next-day courier across Kuwait
- Team
- Founder plus one part-time customer-service hire
A jewellery box full of first orders, and almost no second ones
When we read the customer list by cohort, the picture was consistent across years: a customer bought once, usually as a gift for someone else or for herself around Eid, and then disappeared from the data entirely. Ninety-day repeat purchase was low even by the standards of a category where a necklace is not consumed and reordered like a skincare product. Nobody had ever tried to earn a second sale from her, because nobody had a channel to reach her through beyond an Instagram feed she might or might not see.
The brand had, without realising it, been treating every sale as a one-off transaction rather than the start of a relationship with someone who had just spent real money on something she cared enough about to choose carefully. There was no WhatsApp opt-in anywhere in the purchase journey, no care instructions sent after delivery, and no reason for a customer to think of the brand again until the next Eid campaign happened to reach her.
The pieces themselves gave the brand a natural excuse to reach back out — a tarnish-care reminder, a warranty check-in, a matching-piece suggestion — and every one of those excuses was going unused. The brand had built a gifting business with no mechanism to turn a gift recipient into its own next customer.
What we did — the messages and flows
The losers are here on purpose. A test with only winners was never a test.
WhatsApp opt-in framed around warranty and care, not a newsletter
Winner"Get your warranty confirmation and care tips on WhatsApp" as the checkout opt-in line
- Format:
- Checkout checkbox, repeated on the delivery-confirmation message
Framing the opt-in around a concrete benefit rather than "join our list" gave the customer an immediate reason to say yes, on a product where warranty and tarnish questions are already common. Asking again on the delivery message caught buyers who skimmed the checkout, on the official WhatsApp Business API rather than a personal number.
A care flow sent two weeks after delivery, not on delivery day
Winner"How's the necklace settling in? Here's how to keep it from tarnishing."
- Format:
- One-to-one WhatsApp message, timed two weeks post-delivery
Sent early enough to prevent a tarnish complaint before it happens, and late enough that the customer has actually worn the piece rather than just unboxed it. This flow produced the highest reply rate of any message tested, mostly in plain Kuwaiti dialect, and it is the flow most likely to be forwarded to whoever gave the piece as a gift.
A matching-piece suggestion sent to gift recipients specifically
Winner"Whoever picked this for you has great taste. Here's the bracelet that matches it."
- Format:
- One-to-one WhatsApp message, sent only to orders marked as gifts at checkout
This targeted the exact gap in the business: a gift recipient is a brand-new relationship the brand had never spoken to directly, since the order was placed by someone else. Speaking to her by name about the piece she actually received, rather than the person who bought it, turned a meaningful share of recipients into first-time direct customers of their own.
Sixty and ninety-day lapsed segments, each with one specific reason to return
Winner"It's been a while — the earrings that match your necklace just came back in stock."
- Format:
- One-to-one WhatsApp message, segmented by the specific piece last bought
A specific, product-linked reason to come back outperformed a blanket discount by a wide margin and did not train the list to wait for the next sale. It worked because it referenced the exact piece the customer owned rather than treating her as an anonymous name on a list.
A weekly generic promotional blast to the entire list
Lost"New arrivals this week!" sent to every opted-in customer regardless of what she had bought
- Format:
- Broadcast WhatsApp message, unsegmented, sent every week
This is the loser worth publishing. A generic weekly blast to the whole list produced the highest opt-out rate of anything tested, and it did so quickly, because it treated the same channel that had just been introduced as a trusted, personal warranty line as a discount megaphone. It taught the list to mute the brand faster than it taught anyone to buy.
What we did — the optimizations, in order
Read repeat rate by cohort and by first product bought
We split repeat purchase at thirty, sixty and ninety days by the first item each customer bought, rather than reading one blended repeat rate for the whole store.
Why: This revealed which product was quietly acting as the brand's gateway item, the one whose buyers were most likely to come back, and it was not the bestseller. Reading the number by cohort rather than in aggregate is what made the retention plan specific instead of generic.
Capture WhatsApp opt-in at checkout and again on delivery
Added an opt-in checkbox framed around warranty and care at checkout, and a second prompt on the delivery-confirmation message, both on the official WhatsApp Business API.
Why: Without a channel the customer actually reads, no flow or cadence built afterward has anywhere to land. Capturing the opt-in twice, at the two moments the customer is already paying attention, was what built a usable list in the first place.
Build the flows in Kuwaiti dialect, timed to real moments
Wrote order confirmation, a care and tarnish-prevention message two weeks post-delivery, a matching-piece suggestion for gift recipients, and a review ask timed after the piece had been worn.
Why: A formal Arabic template on a warranty message reads like a bank circular, which undercuts the personal trust the brand needs for a customer to hand over her name for engraving or her preferences for a next piece. Writing in dialect and timing each message to a real moment is what made the flows read as care rather than automation.
Set a segmented cadence instead of a weekly blast
Moved from an unsegmented weekly promotional message to two to four broadcasts a month, segmented by last product bought and by whether the customer was a gift recipient or the original buyer.
Why: The weekly blast was the fastest way to teach the list to mute the brand, and cutting frequency while raising relevance is what turned WhatsApp back into a channel people actually opened rather than one they tolerated.
Target lapsed segments with a specific reason, not a discount
Built sixty and ninety-day lapsed flows referencing the specific piece each customer had bought — a matching earring, a restock, a care reminder — rather than a blanket percentage off.
Why: A blanket discount trains the whole list to wait for the next one rather than buy now, and it says nothing about the customer as an individual. A reason tied to the piece she actually owns performed better and protected margin at the same time.
Speak to the gift recipient, not just the gift buyer
Added a checkout field marking an order as a gift and, where a delivery address or WhatsApp number for the recipient was available, sent her a direct message introducing the care flow rather than only messaging the purchaser.
Why: The gift recipient is the brand's actual next customer far more often than the person who paid, and she had never been spoken to directly before. Opening a channel to her specifically is what let a single Eid gift start a relationship the brand could build on.
Measure repeat rate, revenue per recipient and opt-out weekly, and prune
Tracked repeat purchase rate, revenue per recipient and opt-out rate for every flow weekly, and killed the unsegmented weekly blast as soon as its opt-out rate spiked past the other messages.
Why: A message that sells but drives people to mute the brand is spending down a channel that took real effort to build. Measuring opt-out alongside revenue is what caught the weekly blast quickly enough to protect the list the other flows depended on.
What changed
The table above carries the figures once the client signs them off, and the shape worth flagging is that repeat rate moved without a single storewide discount campaign. It moved because the brand finally had a channel a majority of customers opted into, and because the messages sent through it referenced a real piece and a real moment rather than a generic promotion.
The gift-recipient flow was the most surprising result. It effectively created a second acquisition channel out of orders the brand had already paid to acquire once, since a meaningful share of recipients had never interacted with the brand directly until this message reached them.
The weekly blast that lost is part of the same story. Killing it quickly, once its opt-out rate stood out against every segmented flow, protected the very channel the other five flows were depending on, which is a smaller headline than a repeat-rate chart but arguably the more important decision of the engagement.
What we would do next
Extend the gift-recipient flow to run automatically around Eid and Mother's Day rather than being triggered manually, since those two windows generate most of the gift orders the flow depends on.
Second, build a light referral message into the care flow, asking a satisfied customer to pass a discount to a friend rather than waiting for the next occasion to bring in new buyers on its own.