The scenario
The brand makes majlis seating, sofas and living-room pieces to order, sold direct to Kuwaiti households, usually as part of a new home, a wedding setup or a pre-Ramadan refresh. Every order is a real, personal sales chat: a founder or a salesperson agreeing fabric, dimensions and a delivery date over WhatsApp from a personal business number.
The problem was what happened after the truck left. There was no official WhatsApp channel, only whichever personal number had handled that one order, no message beyond a delivery-day thank-you, and no referral system, even though the majlis is one of the most socially visible purchases a Kuwaiti household makes: guests sit in it and ask where it came from within weeks of delivery.
- Monthly revenue band
- pending client sign-off
- Average order value
- pending client sign-off
- Fulfilment
- Contracted delivery and assembly team, appointments booked directly with the customer
- Team
- Founder, one salesperson, contracted upholsterers
A great first sale, and no channel to reach the customer again
The order chat was genuinely good: the founder answered questions patiently, sent fabric options, agreed a delivery date and followed up with a warm message on the day itself. The problem was that the chat lived on a personal WhatsApp number, mixed in with everything else that number handled, with no tag, no list and no way to message that customer again on purpose six weeks later.
Once a broadcast list did get pulled together, from old order chats copied by hand, a single monthly discount blast went out to the whole list regardless of what anyone had bought or when. It read as spam because, structurally, it was: numbers that had never opted in, on a personal account rather than the official Business API, receiving a generic offer with nothing to do with a wedding-season sofa order from a year earlier. Two numbers were temporarily restricted before anyone traced it back to this list.
Meanwhile the most valuable channel in the business was sitting unused: the guest who sees the majlis in person and asks about it. The brand had a genuinely referable product and no organised way to catch the referral, a working sales channel with no memory, and a broadcast habit that was actively damaging the number it needed for the next order.
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 moved to the delivery and assembly appointment
Winner"Save this number for your delivery updates and fabric care tips" said by the delivery lead on arrival, on the official Business API
- Format:
- In-person opt-in ask at the real moment of delivery, plus a QR code left on the invoice
The order chat happens before trust in the finished product exists. Delivery day, when the customer is looking at the actual majlis in her own home, is where she is most willing to save a number, and asking on the official API rather than a personal one meant the opt-in was finally attached to the business, not to whichever salesperson had handled the order.
A two-week fabric-care check-in
Winner"How's the majlis settling in? Here's how to keep the fabric looking new."
- Format:
- Single WhatsApp text message with a short care-tip image, sent two weeks after delivery
Furniture does not get reviewed or reordered on delivery day; it gets judged after a family has actually lived with it for a couple of weeks. This message had no offer attached and was not meant to sell anything — it was meant to be useful enough to keep the number saved, and it produced the highest reply rate of anything sent.
A referral message triggered right after a five-star review
Winner"Guests always ask where it's from. Send them this and you both get KD 20 off your next piece."
- Format:
- WhatsApp message with a personal referral link, sent only after a positive review
This category's second sale is rarely the same customer buying another sofa — it is the guest who sat on the first one. Naming that exact social moment, rather than offering a generic discount, turned a behaviour the brand already benefited from informally into a tracked, rewarded referral loop.
A generic monthly discount blast to a hand-copied contact list
Lost"20% off everything this month" sent to every number ever contacted, regardless of what they bought
- Format:
- Single broadcast from a personal WhatsApp number, unsegmented
It read as spam because it was structurally unopted-in, unsegmented and sent from a personal account rather than the official API, and it produced complaints rather than orders. Two numbers were briefly restricted, which is a far more expensive outcome than the discount itself, since a restricted number stalls every live order chat riding on it.
A blanket 10% win-back offer to every lapsed quote request
Lost"10% off if you order this month" sent to everyone who had ever asked for a quote and not ordered
- Format:
- One-off WhatsApp broadcast, no segmentation by what was originally asked about
It treated a fabric-availability question, a delivery-timeline question and a simple price-comparison browse as the same lapsed lead, and a flat discount taught the small number of serious buyers on the list to wait for the next blast rather than order when ready. A message tied to the specific thing each person had asked about — a new fabric back in stock, for instance — converted a segment this blast never reached.
What we did — the optimizations, in order
Read the cohort by first product and by referral, not just by repeat
We looked at what each customer's first order was and whether it produced a second sale to the same household, an accessory add-on, or a referral to someone new, since a straight repeat-purchase read undercounts what this category's second sale actually looks like.
Why: A household that buys one majlis rarely buys a second one within a year, so a repeat-rate number alone would have told us this category has no second sale, which is not true — it just looks different from a consumable's.
Capture the channel at delivery, on the official API
We moved the WhatsApp opt-in ask from the personal sales chat to the delivery and assembly appointment, on the official WhatsApp Business API, with a QR code on the invoice as a backup.
Why: A personal number cannot be broadcast from safely and cannot be handed to a new salesperson without losing the history. The official API also gave the brand a template message library, which is what a discount blast should have been using from the start.
Build the flows in Kuwaiti dialect: delivery, care, review, referral
We wrote four short flows — delivery confirmation, a two-week fabric-care check-in, a timed review ask, and a referral message triggered by a positive review — each in Kuwaiti dialect rather than formal Arabic.
Why: A formal Arabic template reads like a bank notice on a channel this brand had only ever used for warm, personal sales conversation, and dialect kept the flows consistent with the tone that had made the first sale in the first place.
Replace the single monthly blast with a segmented cadence
The one generic monthly discount was replaced with two to three sends a month, segmented by season — pre-Ramadan majlis refresh, National Day décor, summer moving — and by what the customer had actually bought.
Why: The unsegmented blast had already cost the brand two restricted numbers. A cadence tied to a season and a purchase reads as relevant rather than random, which is the difference between a message that gets muted and one that gets replied to.
Build the referral offer around the guest, not a discount code alone
The referral message was written around the exact moment a guest asks about the majlis, with a reward for both sides, rather than a generic share-this-code line sent with no context.
Why: The referral was already happening informally every time a guest asked; the brand just had no way to capture or reward it. Naming the moment made the offer feel like a natural extension of a conversation that was going to happen anyway, not a marketing ask.
Win back lapsed quotes with the specific reason they asked
Instead of a flat discount to everyone who had asked for a quote and never ordered, sixty and ninety-day lapsed segments were messaged about the specific fabric, size or timeline they had originally asked about.
Why: A blanket discount had already taught serious buyers to wait for the next one instead of ordering when ready. Tying the win-back to the actual reason someone stalled brings back the person who was close to ordering without training everyone else to delay.
Measure and prune by opt-out, not just by replies
Every flow and broadcast was tracked for opt-out rate alongside reply rate and revenue per recipient, and the segment-wide blast was retired the moment its opt-out rate spiked, regardless of how it had performed on volume.
Why: A message that sells well but pushes people to mute the business number is borrowing against every future order that number was going to carry, and this brand had already lost two numbers to exactly that trade-off once before.
What changed
The table above carries the numbers once the client confirms them, and the shape worth noting is where the revenue came from: not repeat sofa orders, which stayed rare as expected, but referral orders, accessory add-ons and the cross-sell that the two-week care message quietly opened the door to.
The official WhatsApp API changed more than the opt-in rate. Because every flow now lived in one place instead of a personal number, a new salesperson could pick up a returning customer's history instantly, and the brand stopped being one restricted number away from losing its entire order pipeline overnight.
The two broadcasts that failed were the clearest evidence of what this list actually wanted. Neither the generic discount nor the flat win-back produced orders worth the complaints they generated, while every segmented, dialect-written message tied to a real moment — delivery, two weeks in, a five-star review — outperformed them without offering a discount at all.
What we would do next
Build a small accessory and décor-refresh catalogue — cushions, throws, side tables — specifically for the WhatsApp list, since it is a far easier second sale to make than another majlis and it gives the seasonal cadence something concrete to offer beyond a check-in message.
Second, formalise the referral reward into a simple two-sided code with a visible balance, rather than a manually tracked link, so it can scale past the founder personally remembering who referred whom as the list grows past a size one person can track by hand.