What happens to your promotions while your restaurant is closed
A promotion and a storefront are two separate objects on every platform we have examined, and restaurant operators are usually surprised by how independent they are. Closing a branch does not pause its offers. The campaign keeps its own schedule, its eligibility keeps being reassessed, any vouchers already issued to customers stay valid, and on fixed fee placements the meter keeps running against a listing nobody can order from. When the branch comes back, the offer is often the last thing to return, because publication delays are measured in tens of minutes.
Does a promotion switch itself off when the storefront goes dark?
No, and nothing in the promotion tools suggests it should. On Snoonu the campaign builder offers “Fixed Schedule”, “Indefinite Schedule” and “Repeated Schedule” as the three ways to time an offer, and all three are calendar objects with no reference to whether the branch is accepting orders. On Talabat a promotion is a background job against a list of vendors, and the documentation describes it as a processing pipeline rather than something coupled to outlet state. HungerStation’s promotions are built through a separate Partner API from the one that carries store status.
The consequence is simple and easy to miss in a monthly report. A branch that was dark for six hours on a Friday ran a discount for six hours that nobody could redeem, and both facts sit in different systems. Nobody at head office sees them in the same view unless someone deliberately puts them there, which is why a campaign postmortem so often concludes that the offer underperformed.
Which promotion costs keep accruing while nobody can order?
It depends entirely on whether you are paying for time or paying for delivery, and the difference is worth knowing before you buy. Foody’s terms set out two paid placement models side by side. “Recommended Based on Availability” is charged “either per impression (Cost per Mille – CPM) or per click (Cost Per Click – CPC)”, so a dark listing that is not being shown generates no charge. “Recommended Based on Cuisine Filter” is provided “at a fixed fee (Cost Per Placement – CPP)” for a period, and a period continues whether or not you trade. Foody’s own cancellation clause confirms which unit each one counts in, charging a withdrawing partner “proportionally for the days or the impressions they received”.
Careem sits on the delivery side of that line and takes the money from a place worth noting. Its advertising flow states that “the allocated budget will be deducted from your merchant payout”, and its bidding model is cost per click. An outlet that is offline serves no clicks and spends nothing, which sounds like protection until you read the campaign warning next to it: “Campaigns with short flight duration may not deliver in full”, alongside the rule that “End date must fall within the same month as the start date.” A week of downtime inside a two week flight does not extend the flight. It shortens the delivery window against a fixed end date.
What happens to a discount a customer is already holding?
It survives, and the clearest published statement of that is in Foody’s terms for its loyalty coupons. Even where a store leaves the programme entirely, “coupons that have been issued but not used are not lost, but remain valid and can be redeemed by Users until their expiration date”, and where a partnership stops and restarts, unredeemed coupons “will be reissued with a validity of 60 days from their reissuance”. A liability created while you were trading follows you through the closure and out the other side.
That is a specific operational trap for a chain that closes a branch to control cost. The offer stops attracting new customers immediately, because the listing is not visible. The obligation to honour it does not stop, and the customers who arrive holding it arrive at a card that will not take an order. This is one of the few places where downtime and promotion interact in a way a customer notices directly, and it lands as a broken promise rather than as a closed shop.
Does a closed week cost you the promotion for the following month?
On Deliveroo it can, because access to the tools is reassessed on a calendar and judged per site. Marketer eligibility is checked shortly before each month begins against published criteria including “Rejections must be fewer than 8%” and “Star rating must be 3.8 or more”, and a site that misses loses Adverts and Offers “for a full month”. A closure that produced rejections rather than a clean shutdown feeds the first of those directly.
Foody attaches a similar gate to paid placement and adds a tenure test, requiring that a partner “Have an average rating of Users over 3.5” and that “The Partner must be active on the Website / Application for at least 30 days” before it can hold a Recommended Position. Activity, not merely registration. A branch that has spent a stretch of the past month unable to take orders is being measured on a criterion it did not know it was being measured on, and the loss arrives a month after the incident that caused it.
Why is the promotion often not live at the moment the branch reopens?
Because the two things publish at different speeds, and the offer is almost always the slower one. HungerStation states the delay plainly: “The promotion creation can take maximum up to 50 minutes if the request reaches the maximum capacity”, and its own checklist for a promotion that fails to appear ends with the instruction to check that “The sync time (up to 50 min) has passed”. Store status, by contrast, is a single call that takes effect at once.
Talabat adds a quieter version of the same problem. Promotion jobs return per vendor validation results including invalid_vendors, and for an update to an existing promotion the documentation states that “the job status will always be [successful] even if all items from the request are not valid”. A chain that reinstates an offer across forty branches after an incident therefore gets a success message that does not mean forty branches. The reopening is instant, the promotion is not, and the gap between the two is a window in which the branch is trading at full price against a competitor who never closed.
What does an outage do to the campaign report you read afterwards?
It contaminates it, and in a direction that leads to the wrong decision. A campaign that ran across an outage shows fewer redemptions, a lower conversion rate and a worse cost per order, none of which are properties of the offer. If the report is read at brand level, the branch that was dark is invisible inside the average, and the offer looks weaker than it was. If the same offer is then dropped, the chain has taken a pricing decision on availability data it never looked at.
This is the practical reason to hold the two records against the same dates and the same branch. Our July 2026 UAE panel put a typical listing at roughly 9.5 lost trading hours in a month, and the mean interruption on the platforms in that panel ran between about two and three and a half hours depending on the platform. A single such incident inside a fortnight long campaign at a single branch is enough to move that branch’s campaign numbers noticeably while leaving no trace in the campaign tool itself.
What should a chain change about how it schedules promotions?
Three things, all cheap. Start campaigns with enough lead time that the publication delay is spent before trading begins rather than during it, because a 50 minute lag applied at opening is 50 minutes of the day’s best margin. Prefer delivery based pricing to fixed fee placement on any branch with a known availability problem, since CPM and CPC stop charging when the listing stops appearing and CPP does not. And read every campaign result next to the availability record for the same listing and the same dates before drawing a conclusion about the offer.
That last step needs an availability history that is not the platform’s own, because the campaign tool and the status tool belong to the same vendor and neither reports the other. Reading a promotion’s performance against an independent record of when that exact listing was orderable is what Kitchain (kitchain.co) exists to make possible, and the definitions behind those hours are at our methodology.