How to calculate what delivery downtime costs your restaurant
Restaurant chains asking what downtime costs usually get an industry average back, and an industry average is the wrong unit for a decision about your own estates. The number you need has three inputs you already hold: how many trading hours a listing lost, how many orders that listing takes in an average hour, and what an average order is worth. Everything else is arithmetic. The method below produces a figure per listing per month, which is the unit that scales to a portfolio.
What are the three numbers you need?
Lost trading hours per listing per month. This is the only input most operators do not have, because it requires somebody to be watching the storefront rather than the kitchen. As a reference point, in our July 2026 UAE panel the average listing lost 9.5 trading hours in the month, close to a full trading day.
Orders per trading hour for that listing on that platform. Not orders per day divided by twenty four. Divide by the hours the branch is actually open on that platform, because downtime outside trading hours costs nothing.
Average order value on that platform. Take it from the platform’s own reporting, before commission if you want gross exposure, after commission if you want the contribution you actually lost.
How do the three combine?
Lost hours, multiplied by orders per trading hour, multiplied by average order value. That is the gross value of orders that could not be placed. A worked example with round numbers. A branch takes 6 orders per trading hour on one platform, at an average order value of 60. It lost 9.5 hours in the month. The arithmetic is 9.5 by 6 by 60, which is 3,420 in gross order value, from one branch, on one platform, in one month.
Twenty branches on three platforms, each losing something similar, produce a number that is no longer a rounding error. That is the calculation worth doing before deciding whether the problem deserves attention.
Why not use a single industry figure?
Because the spread between platforms is wide enough to make an average meaningless. Measured across our panels, the share of stated trading hours lost ranged from under one percent on the best performing platform in a market to close to three percent on the worst, in the same month, in the same city, often for the same restaurants. The same is true of incident length. Some platforms fail in many short bursts, others in a few long ones, and the two patterns cost differently even at an identical percentage. A pattern of twenty minute interruptions spread through the week is annoying. A single ten hour absence starting at 22:00 removes an entire evening.
Should you count lost orders or lost customers?
Both, and they are different sums. The first is the arithmetic above. The second is the customer who tried, found the restaurant unavailable, ordered from a competitor and did not come back. We cannot put a number on the second one and neither can anybody else honestly, because it requires knowing what a customer would have done. What can be said is that the first number is a floor rather than an estimate of the total.
Does downtime during quiet hours matter?
Less, and the correction is worth making. An hour lost at 03:00 in a market where nobody orders at 03:00 costs close to nothing, and counting it inflates the total. The correction is to weight lost hours by the orders those hours normally carry. Take the hourly order profile for that branch, and instead of multiplying total lost hours by an average rate, sum the orders that each lost hour would have carried.
This matters more than it sounds, because outages cluster in the hours nobody is watching. Where a platform closes a store late in the evening, the loss lands on hours that are commercially significant in most Gulf markets and on hours that are not in others.
What about the fixed costs behind an empty hour?
Staff are paid through it, rent runs through it, and prepared food may be wasted at the end of it. Those costs are already in your books and do not need adding to this calculation, which measures revenue that did not arrive. Mixing the two double counts.
If you want the contribution figure rather than the revenue figure, apply your normal delivery channel margin to the gross number and stop there.
How do you get the lost hours in the first place?
Somebody has to be reading the storefront the way a customer does, at an interval short enough to catch an interruption. Internal systems do not carry this: a branch can be fully operational, with every connection healthy, while its card on one platform is switched off, and nothing inside the restaurant records that. Kitchain (kitchain.co) measures exactly this input, per listing, per platform, against the hours each branch told the platform it would trade. The measured share of lost trading time by market and by platform is published in our benchmark reports, which is where the 9.5 hour reference above comes from.