The scenario
The brand sells bottles, feeding sets and bath gear in Kuwait, on a Shopify store with KNET, Tabby and cash on delivery. Two years in, no in-house retention hire, and a product range that had become a genuine baby-shower favourite — the kind of gift multiple guests would buy for the same expecting friend after seeing it recommended in a WhatsApp group.
That popularity made the first order easy and disguised a real gap: once a customer received the product as a gift or bought it once for a newborn, nothing in the business ever spoke to her again except an occasional, unsegmented email discount. The founder had built a brand people loved to gift. She had never built a reason for the person using it to come back and buy from the brand directly.
- Monthly revenue band
- pending client sign-off
- Average order value
- pending client sign-off
- Fulfilment
- Own stock, local courier, next-day across Kuwait
- Team
- One founder, one customer service hire answering WhatsApp and email
The favourite gift nobody could sell twice
The cohort numbers told a clean story once we split them by first product bought: customers who started with the feeding set had the strongest thirty-day return of anyone in the file, far ahead of any other line — a clear gateway product hiding in plain sight. Nobody had ever looked, because the brand's only retention activity was a monthly discount email sent to everyone regardless of what they had bought or when.
The gifting pattern made the gap worse rather than better. A large share of first orders were bought by a friend or relative for someone else, which meant the person actually using the product every day, the one with every reason to reorder or refer, had frequently never given the brand her own contact details at all.
The brand did not have a product problem or a demand problem. It had never captured the one channel — WhatsApp — that the person actually using its products already lived in, and every order since launch had ended the relationship the moment it shipped.
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 at checkout and again at delivery
Winner"Want your delivery updates and reorder reminders on WhatsApp instead?"
- Format:
- A checkbox at checkout on the official WhatsApp Business API, reconfirmed on the delivery-confirmation message
This was the single fix the brand needed most: a channel the actual product user reads, captured at the two moments she is already paying attention — placing the order and receiving it. It converted at a far higher rate than the old email signup ever had, because it asked for a channel she checks daily rather than one she checks once a week.
The stage-up nudge, timed to the gateway product
Winner"Growing fast? Here's what most parents add next at this stage."
- Format:
- WhatsApp template message sent at the feeding set's typical replacement interval, written in Kuwaiti dialect
Timed to when a family naturally moves up a bottle or cup size rather than sent on a fixed calendar date, this flow read as a helpful reminder rather than a sales push, because it matched a change every household in this category goes through on its own schedule regardless of the brand.
The blanket monthly broadcast to the whole list
Lost"This month's discount, for everyone on our list."
- Format:
- A single WhatsApp broadcast sent to every opted-in contact regardless of purchase history
This was the brand's original habit carried over from email, and it performed worse on WhatsApp than it ever had by email: opt-out rate spiked noticeably every time it went out, because a channel this personal punishes an irrelevant message far more than an inbox does. It is exactly why segmentation by last product had to replace it rather than sit alongside it.
The gift-giver referral, sent from the gateway cohort
Winner"Know someone expecting? Send her this, you both get something."
- Format:
- WhatsApp message sent only to customers whose first purchase was the feeding set, with a shareable referral code
Targeted only at the cohort that already had the strongest early return, this flow made explicit what was already happening informally in baby-shower groups, and it converted well because it gave a name to an act — recommending a gift — the recipient was already inclined to do anyway.
A blanket sixty-day lapsed discount
Neutral"We miss you. Here's a discount to come back."
- Format:
- WhatsApp message to every customer with no order in sixty days, same offer regardless of what they had bought
Adequate but unremarkable. A discount with no stated reason performed no better than doing nothing for customers whose child had simply outgrown the product's original stage, because the offer never acknowledged why she had gone quiet, and a specific reason to come back consistently outperformed a generic one in every later test.
What we did — the optimizations, in order
Read the cohorts by first product bought
We split every customer by the first item they ever bought and tracked repeat purchase at thirty, sixty and ninety days per group, which surfaced the feeding set as the clear gateway product.
Why: Without this split every customer looks the same in an account-wide average, and the brand had been running one blanket offer to a list that actually contained a small group of eager repeat buyers hidden inside a much larger group who had only ever received a gift once.
Capture WhatsApp at the two moments attention already exists
An opt-in checkbox went live at checkout on the official WhatsApp Business API, reconfirmed on the order and delivery-confirmation messages rather than left as a single ask.
Why: The gifting pattern meant the person placing the order and the person actually using the product were often different people, so capturing the opt-in only once, at purchase, still risked losing the one contact detail the brand actually needed.
Build flows around the stage change, not the calendar
Order confirmation, delivery update, then a stage-up nudge timed to the gateway product's typical replacement interval, then a review ask and a cross-sell into the next item most households add at that point.
Why: A message tied to a real, predictable change in what a growing family needs reads as useful information rather than a sales attempt, which is the difference between a flow customers keep and one they mute after the first send.
Segment every broadcast by cohort, never send to the whole list
The blanket monthly discount was retired and replaced with two to four broadcasts a month, each segmented by first product and time since last order, after the unsegmented version spiked opt-outs.
Why: WhatsApp punishes an irrelevant message harder than email does, and the opt-out spike on the blanket broadcast was the clearest signal in the whole engagement that segmentation was not optional on this channel.
Turn the gift-giving pattern into a referral offer
A referral message with a shareable code went out specifically to the gateway-product cohort, giving a name and a small incentive to a recommendation they were already making inside their own WhatsApp groups.
Why: The brand's real growth engine had always been word of mouth inside mothers' groups; the referral flow simply gave that existing behaviour a code to track and a reason to happen a little more often, rather than trying to invent a new channel from nothing.
Give the win-back a reason, and measure what it costs to keep
The generic sixty-day discount was replaced with a message naming the specific reason a customer had likely gone quiet — a child aging past the gateway product's stage — paired with the item that actually fits that new stage, and every flow's opt-out rate was reviewed monthly against its revenue per recipient.
Why: A discount with no stated reason applies the same offer to every lapsed customer, and the ninety-day cohort read had already shown that the reason for going quiet is usually predictable and specific, which means the win-back message can be too.
What changed
The headline number sits in the table above, and the shape of it matters: repeat purchase rose specifically among the cohort that started with the gateway product, which is exactly where the cohort read said the opportunity was concentrated, rather than lifting evenly and unexplainably across the whole list.
WhatsApp opt-in became the far larger channel within a few broadcast cycles, and revenue per recipient on the segmented sends held well above the old blanket email discount, while opt-out rate on the same segmented flows stayed low precisely because nobody outside the relevant cohort ever received them.
The clearest instructive result was negative: the single blanket broadcast, carried over from the old email habit, is the one thing in this engagement that made a metric move the wrong way, and retiring it in favour of segmentation was worth more than any single new flow that replaced it.
What we would do next
Add a saved gift-recipient field at checkout so a gift-giver can pass the delivery and stage-up flows directly to the new mother, closing the exact gap that let the original contact go missing at the point of purchase.
Extend the referral offer to the second-strongest cohort now that the gateway-product version has a season of results behind it, and test whether the same specific-reason approach lifts the ninety-day lapsed segment as much as it lifted the sixty-day one.