Commission on loyalty
Regulars ordering the same dishes weekly through aggregators — a percentage of the group’s most reliable revenue leaving with every order.
Case StudyRestaurants & F&B
A Dubai Restaurant Group: Taking Orders Back From the Apps
A family-run restaurant group in Dubai watched delivery aggregators take a heavy commission on orders from customers who were already loyal. We built its commission-free ordering channel and a bilingual WhatsApp AI that takes orders through the Friday rush — keeping the apps for discovery and the margin for the family.
At a Glance
- Client
- Family restaurant group — Dubai
- Industry
- Restaurants & F&B
- Engagement
- Direct ordering + WhatsApp AI
- Languages
- Arabic & English
- Core target
- Repeat orders direct, rush answered
Background
The Context That Made This Worth Building
The group’s food had built its reputation over a generation; the aggregator apps had built their toll booth in five years. Every repeat customer ordering through an app cost the family a commission on a relationship they had already earned — and during service rushes, the phone rang unanswered while the kitchen worked flat out.
The Challenge
The Pain Points That Started It
Two leaks, one at each end of the order.
The rush answers no phones
Friday evening is peak demand and peak kitchen load simultaneously; calls and messages went unanswered exactly when they were most valuable.
No data on their own customers
Aggregators keep the customer relationship. The group couldn’t message its own regulars about new dishes or Ramadan hours.
Discovery
Understanding Before Building
Order history analysis (from aggregator statements and till data) showed the classic split: discovery orders from new customers, and a thick band of repeat orders from regulars — the band worth moving to a direct channel. We modelled the margin difference and set the target: shift a meaningful share of repeat orders direct within months.
Observing a Friday service settled the second design question. No one can staff the phone at 8pm on Friday — so the channel that answers must not need staff. The WhatsApp AI had to handle the menu, options, totals, and handoff to the kitchen flow without a human in the loop.
The Solution
What We Designed and Built
A direct channel regulars prefer, and an AI that never misses the rush.
Commission-free ordering
Branded ordering site — full menu, options, cards and cash on delivery — feeding orders straight into the kitchen flow.
WhatsApp AI order-taker
Conversational ordering in Arabic and English: menu questions, customisations, totals, and payment preference — unlimited simultaneous conversations at peak.
Regulars & loyalty layer
Customer history and consent-based messaging — new-dish announcements, Ramadan timings, and a simple perk for ordering direct.
Scope Delivered
Everything That Shipped
- Branded ordering website (AR/EN)
- WhatsApp AI taking orders end to end
- Kitchen order flow integration
- Cards + cash-on-delivery payments
- Customer history & loyalty perks
- Broadcast messaging to consenting regulars
- Menu management across branches
Implementation
How the Build Ran
- 1
Order economicsWeek 1
Aggregator statements analysed, repeat band sized, direct-channel target agreed with the family.
- 2
Ordering channelWeeks 2–4
Ordering site and kitchen flow built; menu structured once, reused everywhere.
- 3
WhatsApp AIWeeks 4–6
Assistant trained on the real menu and real customer phrasing — including mixed Arabic-English orders — then stress-tested against rush-hour volumes.
- 4
Shift the regularsWeeks 6–10
Table cards, box inserts, and a direct-order perk moved the habit; aggregators kept for discovery.
Stack
- Next.js
- Node.js
- WhatsApp Business API
- OpenAI
- Stripe
Results & ROI
What It Delivered
0%Commission on direct orders
100%Rush-hour messages answered
~30%Repeat orders shifted direct in months
2Languages the AI orders in
- The loyal-customer band orders direct — margin restored on the revenue the family had already earned
- Friday rush enquiries answered instantly and completely, with the kitchen untouched
- The group finally owns its customer list — and can speak to it
- Aggregators repositioned as paid discovery, not the whole business
Nothing about the food changed. The group simply stopped paying rent on relationships it owned. The AI made the direct channel feel effortless at exactly the hours it used to be impossible, and the loyalty layer gave regulars a reason to make direct ordering the habit — the compounding kind of fix that shows up in monthly profit, forever.
Takeaways
What This Project Teaches
Move the band, not the business
The goal was never leaving the apps — it was shifting the repeat band where commission hurts most. Precision beats ideology.
Design for the rush, not the average
A channel that fails at peak demand fails at its whole job. The AI was stress-tested against Friday, not Tuesday.
Habits need a nudge with a reason
Table cards plus a small direct-order perk did more than any banner — habit change is physical and incentivised.
Before You Ask
What Readers Usually Want to Know
Do we have to leave the delivery apps?
No, and we would advise against it. The apps are paid discovery — they introduce customers you would not otherwise reach. The goal is to move the repeat band, your regulars ordering the same dishes weekly, onto a direct channel where you keep the margin and the customer relationship. Most healthy groups run both deliberately.
Does the ordering system work with our POS and kitchen flow?
That is the integration that decides whether staff actually use it. Orders should reach the kitchen the same way every other order does — printer, screen, or POS — rather than arriving as a separate stream someone has to re-key during service. Where a POS exposes an interface we connect to it; where it does not, we design the flow around how your pass actually works.
What happens when the AI cannot handle an order?
It hands over to a human with the full conversation attached, and the customer is told plainly that someone is joining. Unusual requests, complaints, and large or bespoke orders are exactly the moments a person should take over — the assistant exists to handle the routine volume that used to go unanswered during service, not to replace hospitality.
The Services Behind This Build
Need the Same Capabilities?
Want a Result Like This?
Tell us your pain points — we’ll walk you through how we’d scope it, what it would cost, and what a realistic outcome looks like. Honestly, both ways.