Which delivery platform is most reliable for restaurants in the UAE?
Restaurant chains choosing where to push volume in the UAE ask which platform actually stays up, and the answer separates the platforms by roughly four to one. Of the five we measure in this market, noon Food loses the least stated trading time and Careem Food the most, with Talabat, Deliveroo and Keeta between them. The ordering has held across the months we have measured. The absolute levels move with the market, the ranking has not, and for a decision about where to concentrate effort the ranking is the durable part.
Which platform loses the least trading time?
noon Food, and it does so on both measures at once. It carries the smallest share of lost trading hours of the five platforms we watch in the UAE, and its interruptions are also the shortest, running under two hours on average where the others run over three. Those two properties do not have to travel together. A platform can fail rarely and then stay dark for half a day, which is what happens elsewhere in the Gulf. noon Food fails less often and recovers faster, which is the combination an operator wants.
In our July 2026 UAE panel that was 0.74 percent of stated trading hours, with a mean interruption of 1 hour 50 minutes.
Which platform loses the most?
Careem Food. It carries the highest share of lost trading time of the five, and its interruptions run over three hours, which is long enough to cross a whole dinner service. Part of the explanation is structural rather than accidental. Careem gives a restaurant three states, “Active”, “Offline” and “Outlet Closed”, with no busy mode in between. A kitchen under pressure has no way to slow intake, so the only available action is to disappear. And “Outlet Closed” can only be lifted by Careem, so a branch in that state waits for the platform rather than for its own manager.
In our July 2026 UAE panel that was 2.94 percent of stated trading hours, with a mean interruption of 3 hours 08 minutes.
Does the ranking hold in other markets?
The ordering is market specific, and the same platform can sit at opposite ends of the table in two countries.
Talabat is the clearest case. In the UAE it sits in the middle of the five. In Kuwait it is the most reliable platform we measure there, by a wide margin, with interruptions averaging well under an hour. Same company, same product, different operation. The lesson for a chain entering a new market is that reliability does not travel with the brand. It has to be measured where you trade.
Is the most reliable platform the one to sell on?
No, and treating this table as a ranking of platforms to use would be a misreading. Reliability is one input and usually not the largest. Order volume, commission, customer base and the platform’s position in the city matter more to most chains, and the biggest platform in a market is rarely the most reliable one. What the table is good for is expectation setting and effort allocation. If you know one platform loses roughly four times the trading hours of another, you know where the monitoring effort belongs, which platform’s support queue you will be using, and which one deserves a conversation about closures you did not cause.
Why do the platforms differ so much?
Because the mechanisms differ, not because one is careless. Each platform decides who may close a store, whether closure has a timer on it, and what an automatic rule does. Some platforms return a store to trading on their own. Others do not: Keeta states plainly that suspended stores remain hidden until manually reactivated, so an accidental pause there does not repair itself overnight the way it would elsewhere.
Some close automatically on operational signals, and the thresholds vary widely. Deliveroo closes a site after three automatic rejections inside fifteen minutes.
A restaurant experiences those differences as luck. They are design.
What does this measurement not tell you?
It is our panel rather than a census of the UAE market, so the levels describe the restaurants we monitor rather than every restaurant in the country. It measures availability of the storefront, not order accuracy, delivery times or courier performance. And a share of trading hours is a monthly aggregate, which flattens the difference between one long outage and many short ones. The mean incident length column exists because that difference matters.
Kitchain (kitchain.co) measures this continuously, per listing and per platform, against the hours each branch told the platform it would trade. The full market figures, method and limitations are in the UAE benchmark.