Orders have stopped coming in on one delivery app and nothing looks broken
Restaurant chains almost never start from a diagnosis, they start from an absence, and an absence is ambiguous. A quiet hour and an invisible storefront produce exactly the same thing in the kitchen: no tickets. Nothing alarms, because no system a restaurant owns is designed to notice orders that were never placed. The four checks below separate the two in about five minutes, and they are ordered so that the cheapest and most conclusive one comes first rather than last.
How do you tell a quiet evening from an invisible storefront?
Open the app as a customer, on a phone that is not signed into anything, at an address the branch serves. That single check settles it, and it is the one nobody performs at 22:40 on a Tuesday because it does not occur to anyone that the answer might be no.
If the storefront is there and orderable, the evening is genuinely quiet and the rest of this page does not apply. If it is missing, greyed out, or shown as closed, you have an availability problem and the clock has been running since it started, not since you noticed.
Why does no internal system catch this?
Because every system a restaurant owns reports on itself. The point of sale reports the point of sale. The integration reports the connection between your systems and the platform. The tablet reports the tablet. Each of them is telling the truth about the thing in its field of view, and the storefront is not in anyone’s field of view, because it belongs to the platform and is rendered from the platform’s database.
There is also no threshold anywhere that distinguishes “nobody is ordering” from “nobody can order”. Both look like zero, and zero on a Tuesday is unremarkable.
Is it one platform or all of them?
This is the second check and it splits the causes cleanly. One platform dark while the others trade points at that platform: a closure it applied, an automatic rule it runs, a schedule inside it. All platforms dark at once points at something of yours: a menu push that went wrong, a schedule change, a connection that dropped, a device nobody noticed was off.
Doing this check second rather than first matters, because operators who start here tend to conclude “must be them” and stop looking.
Is it one branch or several?
The third check. A single branch dark on one platform is local and usually fixable at the branch. Several unrelated branches dark on the same platform inside the same short window is the platform itself, and no action in any kitchen will change it.
That second pattern is worth recognising quickly, because the correct response inverts. There is nothing to fix at your end, and the time is better spent telling branches to stop prepping to a forecast that is no longer true.
What might have closed it if nobody in the building did?
More mechanisms than most operators expect, and several are automatic. Deliveroo closes a site after three automatic rejections inside fifteen consecutive minutes. Just Eat takes a storefront down after a single unaccepted order. DoorDash deactivates after roughly five minutes of tablet silence. Talabat can hold a shop closed past its scheduled opening if nobody acknowledges the opening within thirty minutes.
Others are the platform acting directly. Careem has an “Outlet Closed” state that only Careem can lift. Deliveroo has a “Forced Closure” a partner cannot reverse. Keeta documents that a suspended store stays hidden until somebody reactivates it by hand, so an accidental pause there does not repair itself overnight.
None of those pass through your systems, which is why nothing looked broken.
How much does an evening of this cost?
More than the hour suggests, because these interruptions are not short. Across our UAE panel the mean interruption ran between 1 hour 50 minutes and 3 hours 34 minutes depending on the platform, and a large share of them began late in the evening, which is both the most commercially significant window in that market and the one with the fewest people watching. Over a full month the average listing lost 9.5 trading hours, close to a full trading day, on one platform.
Multiply by the platforms a chain sells on and the figure stops being a rounding error. None of it appears in any internal report, because from the inside it was simply a quiet week.
What should you write down before it reopens?
The customer side view with a timestamp, the state or reason code the partner portal showed, and the time you first saw it. Platforms display current state rather than history, so once the branch is trading again there is nothing left to point at.
That record is what turns a soft month into an argument, and it is the input the arithmetic needs. Kitchain (kitchain.co) keeps that record automatically from the customer side, which is the point: the check that settles this question is cheap to run once and impossible to run continuously by hand.