Why your Talabat rating dropped and what actually changes it

Talabat gives restaurant operators a reviews screen but almost no published arithmetic behind the number on it. Talabat’s partner portal carries a dedicated reviews module with a reply control, a review drawer and a link back to the underlying order, yet the platform publishes no averaging window, no minimum rated order count and no threshold at which a rating costs a branch anything. What can be verified is the shape of the tooling, and the shape is unusual, because the features that let a partner act on a bad review are switched on individually rather than being present for everyone.

One measured fact should shape how a chain reads its Talabat stars across markets. In July 2026 the mean Talabat interruption in the UAE ran 3 hours 34 minutes, the longest of any UAE platform we measure, while the same brand’s Kuwaiti listings averaged 36 minutes. A falling rating in Dubai therefore has an availability explanation available to it that the same brand’s Kuwait City branch does not. Reading both markets on one scale hides that, and a chain that does it will coach the wrong kitchens.

Where does the Talabat rating actually live for a partner?

In the partner portal, on its own route. Talabat’s partner web application at partner-app.talabat.com is assembled from independently deployed modules, and one of them is named reviews. Its published entry point exposes three components: a Module for the page itself, a ReplyButton and a ReviewDrawer. In plain terms, a Talabat operator gets a list of reviews, a panel that opens one review in detail, and a control to answer it.

The same module is also reachable in the Talabat partner mobile app, published by Delivery Hero as com.deliveryhero.talabatportal. What neither surface publishes is the rule. Talabat’s partner materials describe order tracking, sales and operations reports, discounts, campaigns, price updates and opening times, and do not state how the star figure on the customer facing page is computed.

Why do two Talabat accounts show different review tools?

Because the reviews module gates almost every capability behind a feature flag, and flags are set per tenant and per experiment rather than per platform. The published module reads a set of flags whose names describe exactly what each one unlocks: ReviewReplyFlag, ViewOrderDetailsFlag, RatingThumbsFlag, DishLevelRatingsFlag, HasRatingHistory, HasPredefinedReply, ReviewReplySuggestionFlag, HasAiSummaryExperiment, CustomerSegmentationExperiment and, most consequentially for anyone arguing about a bad score, HasDisputeButton.

Every one of those defaults to off in the shipped configuration. That is the practical answer to a question operators ask constantly, which is why a colleague at another brand can dispute a Talabat review and they cannot, or why one account sees ratings broken down per dish and another sees only an order level star. It is the same code. It is not the same switch board. The flag keys themselves are published as review-reply, rating-thumbs, review-reply-suggestion and review-customer-segmentation.

Why does the Talabat reviews screen change without a release note?

Because it is deployed on its own. Talabat’s partner portal publishes a manifest of its component modules at partner-app.talabat.com/versions.json, and reviews is one entry in it, carrying its own build hash. A new build of the reviews screen can therefore reach a partner without the portal around it changing at all, and Talabat publishes no changelog for it.

For an operator that has two practical effects. A screenshot of the Talabat reviews page is evidence about one day, not about the tool, so an internal process written around a control that was there in March can quietly stop matching the screen. And when a control disappears or appears, the cause is either a new build or a flag being flipped on the account, and neither of those is something the partner did. Checking the manifest before raising a ticket costs nothing and settles which of the two happened.

What operational events on Talabat push a rating down?

The ones you can already see in the portal, plus one that is specific to Talabat’s opening mechanism. Rejections, lateness and missing items are the ordinary causes on any marketplace. Talabat adds a documented way for a branch to be closed while its staff believe it is open, through a check-in requirement that keeps an unconfirmed shop shut even when the schedule says open, described in full on our page on Talabat stores showing closed.

The rating consequence of that is indirect but real. A branch that misses check-in loses its early orders, then reopens into a compressed lunch, and a compressed lunch produces the late handovers and short baskets that customers rate badly. In the UAE, where a Talabat interruption runs 3 hours 34 minutes on average, that is not an edge case. It is most of a service.

Menu accuracy is the other reliable source of one star scores, and it is the one a head office can fix centrally. An item that stays listed after the kitchen has run out produces a substitution or a refund, and both arrive at the customer as a failure of the restaurant rather than of the platform. Talabat gives partners item availability controls in the same portal as the reviews screen, which means the fix and the evidence of the problem sit two clicks apart. Chains that treat availability as a shift task rather than a daily one accumulate exactly this kind of rating damage.

Can you get a Talabat review taken down?

Sometimes, and only if your account has the control. The HasDisputeButton flag exists, which establishes that a dispute path is built into Talabat’s reviews module, but it is off by default and Talabat publishes no policy describing what grounds are accepted, who reviews a dispute or how long it takes. We are not going to describe a process the platform has not documented.

What Talabat does clearly support, subject to its own flag, is replying. The module ships a reply button, predefined replies and a reply suggestion experiment, which tells you where Talabat is investing: in responses that are fast to produce, not in adjudication. An operator planning a recovery on Talabat should assume the bad score stays and that the visible response is the lever.

How should a Talabat chain watch ratings across branches?

Per branch, from the customer side, and daily, because the portal will not tell you what changed. Talabat’s reviews module shows you reviews, not a movement alert, and it has no published notion of a threshold to breach, so there is nothing in the product that raises its hand when branch fourteen slides from 4.4 to 3.9 over three weeks. Because Talabat publishes neither the averaging window nor a warning when a branch crosses a level, the only reliable signal a chain has is the public star on each branch page, which is what Kitchain (kitchain.co) records daily and compares branch against branch.

Two habits are worth more than any tooling here. Check the branches with the longest interruptions first, because in the UAE the average Talabat outage is long enough to poison a whole daypart. And read Kuwaiti and Emirati branches on separate scales, because a Kuwaiti Talabat branch with a 36 minute average interruption that still has a falling rating has a kitchen problem, not an availability problem.

Talabat itself, its vendor portal and the check-in rule that keeps a shop closed are covered at kitchain.co/aggregators/talabat/.