Margin Intelligence: where 24.3 points of GMV go
In Bahrain, the UAE, Qatar, Saudi Arabia, Oman and Kuwait, everyone is asking who wins the ordering screen once an AI assistant starts placing orders, and as management consultants we call that batting average. Margin Intelligence asks where margin is created, and who captures it.
What talabat's marketplace keeps, as a share of everything customers order
24.3 points
talabat's Q2 2026 results show what the marketplace keeps as a share of GMV. Commission, delivery fees, service fees, subscriptions and ads, less vouchers, add up to 24.3 points.
Source: talabat Q2 2026 results, released 12 August 2026.
How to read the 24.3 points
Commission is the largest line at 13.1 points, a little over half the total. Delivery and service fees add a further 8.5 points. Ads, at 3.4 points, are the line most plainly priced to a person scanning a list, which is why the sponsored list is where the agent question began.
A restaurant or brand that buys the whole stack as one bundle cannot negotiate it line by line. Splitting it shows which lines would change if the customer arrived by another route, and which would be paid whoever sent the order.
Which points are in play
At Par attributes 8 or 9 of those 24.3 points to winning the customer: vouchers, marketing and profit. The rest pays for what any order needs, whoever sends it: the rider, payments, refunds and support.
If an assistant starts sending the orders, only the first slice is in play. We size that exposure from inside, on your own orders.
The rest stays, and almost nobody reports it. Gulf platform disclosures leave out orders per rider hour and what a refund costs per order.
Batting average and getting on base
Baseball teams once paid for batting average and underpaid for getting on base. Gulf commerce has its own batting average in GMV and order growth. What an order costs when it arrives right sits in the same blind spot, only because nobody publishes it.
The line we would put on a board page is cost per order that arrived right, refunds and re-deliveries included. A group that cannot produce it cannot say which of its points are winning customers and which are serving them.
The board's questions
- A Tenant should ask how much of what it pays a platform buys customers, and how much pays for rider, payment and support work it would otherwise carry itself.
- A Carrier should ask what share of a platform's points reaches the fleet, and whether its rate improves when orders arrive right.
- A Merchant should ask what it would keep if its own channel replaced the first slice, and what it would then pay for the rest.
- A Sovereign should ask whether management can produce cost per order that arrived right today, and by channel.
Layer 6 for your group
We build the cost-per-right-order view for your group, divide margin between winning the customer and serving the order, and test it against agent scenarios on the Moneyball Finance method. The aggregator margin tool gives a restaurant group its own figures, and every engagement starts with the Control Diagnostic.
Sources
- talabat, Q2 2026 results, released 12 August 2026: commission 13.1%, delivery fees 6.2%, service fees 2.3%, subscriptions 1%, ads 3.4%, less 1.7% in vouchers, totalling 24.3 points of GMV.
- The split of the 24.3 points between winning the customer (8 or 9 points) and serving the order: At Par analysis. Published in Moneyball Finance, 9 October 2026.
Related
- The checkout bill for agent-led commerceCard fees stay put when an agent pays. The new cost is a charge for access to the customer, and you need to know whether it buys new orders or taxes existing ones.
- Moneyball Finance: our methodOur method: find the measures that create value and the headline numbers that mislead.
- Restaurant aggregator marginThe contribution a restaurant keeps from one aggregator order, and what it adds up to in a month.
Bring last month. Leave with a score.
Sit down with Qasim Abbas. One table, your numbers, and a clear read on where you stand against par.