How Many Hours a Month Is Your Restaurant Invisible?

Three and a half hours. That is the average time a restaurant spends offline on delivery apps every month, measured by Delaget across more than 30,000 locations, at $24 of lost revenue per hour. The worst performers sit at 58 hours. For them it works out to $1,392 a month, close to $17,000 a year, per store.

Here is the part that should bother you. When a register dies, six people start shouting within ninety seconds. When your store drops off DoorDash, nothing happens. The lights are on, the cooks are standing there, and the tickets just stop. It looks like a slow Tuesday.

It is not a slow Tuesday.
‍

‍Do the math on your own group before you read further. Ten locations at the poor-performer average means $170,000 gone in a year. That is the annual profit of one restaurant, spent on nothing at all.

The exposure grew because the channel grew. Delivery made up 2.4% of quick-service revenue in 2020. By 2023 it was 11.1%, according to Delaget's data. So the hours you cannot see are hours inside the channel carrying a ninth of your sales.

Deliverect gets to a similar place from the other direction: a 100-location brand averages 5% downtime during its own stated open hours. Across the estate that adds up to 322 hours a week and roughly £50,000 in weekly lost revenue.

And there is a cost nobody puts in the spreadsheet. During an outage the platform reroutes orders to a nearby restaurant, and it does not route them back until your issue clears. You are not pausing sales. You are handing them to the place two blocks over, and a chunk of those customers find they like it there.

Deliverect splits restaurant downtime in two, and the split decides what you can fix.

Desired downtime is a decision. The walk-in died, the line is forty tickets deep, two cooks called out, so a manager pauses the store. Correct call. Someone chose it.

Undesired downtime is everything else. A tablet that lost Wi-Fi. A POS integration that quietly stopped passing orders. A platform that deactivated the store over metrics nobody was watching. Nobody chose any of it, and nobody knows.

Most operators look at one combined number and conclude they have a discipline problem with the staff. The split usually says otherwise. In Delaget's breakdown of outage causes, intentional pauses account for 10%. The smallest category. Also the only one that ever comes up at the manager meeting, because it is the only one with a human being to blame.

The biggest category is the platform itself, and the threshold is lower than almost anyone assumes. DoorDash defines downtime as the share of your scheduled open hours when the store is unavailable, and sets the target at 1% or lower. Across a 70-hour week that is 42 minutes. Worth sitting with for a second: 42 minutes of cumulative failure in a week, including every tablet hiccup and every forgotten pause, and you are at goal.

The goal itself comes from the top 20% of merchants. DoorDash builds its Operations Quality targets off their average annual performance, so you are not being measured against the industry. You are being measured against the best fifth of it.

Deactivation is mechanical. Two or more avoidable cancellations in one day can take your store down. What counts as avoidable: hours that do not match reality, an item still showing in stock after it ran out, a technical error on the tablet or POS. Keep that rate under 1.1% and you stay eligible for Most Loved. Go over, and the platform decides your availability instead of you.

Delaget's numbers say this is the main event. Seventy percent of outages begin with the platform closing a store over reported problems: driver wait times, missing items, cold food. Another 20% come from driver reports, usually a driver arriving to a locked door because the listed hours run fifteen minutes past the real ones.

Then there is the tablet, the most fragile object in your building, living where nobody looks. DoorDash lists the states without any drama: offline, uncharged, or Order Manager closed. Each one makes you invisible to every customer in the delivery radius. A tablet lying face down on a shelf behind the pass with a dead battery has the same commercial effect as bolting the front door at noon. It just does not feel that way, because nobody walks past a bolted door and shrugs.

Integration errors are rarer and much nastier. Delaget found them hardest to diagnose: an expired login, a weak Wi-Fi signal, and in one real case a bad cable. Orders leave the app and never arrive at the POS. The platform still shows you as open. The kitchen gets silence. Everyone assumes business is slow.

Uber Eats runs a store status endpoint that flips a location between online and "currently unavailable." Useful when it works. The flip side is that every piece of middleware in your stack can close your restaurant without a human touching anything.

Yes, and for a structural reason: a virtual brand has no door.

A physical restaurant that goes dark on DoorDash still serves walk-ins and still runs a drive-thru. The dining room absorbs part of the hit. A delivery-only brand offline on the apps is a business that does not exist until someone fixes it. No storefront, no discovery, no revenue, and no second channel to fall back on.

Multi-brand kitchens compound it. One kitchen running four concepts carries four listings, four sets of hours, four tablets or four integration paths, and four separate ways to break. A single POS error kills all four at once. The operator usually learns about it from the weekly report.

Ranking makes the hour cost more than an hour. Platforms weight availability in what customers see first, so patchy uptime costs you placement, placement costs you the orders it was generating, and getting the position back takes sustained good behavior rather than one good night.

Stop reading reports. Start getting alerts. Retrospective data is a receipt, not a tool.

Burger King UK is the clearest case I have seen. Across 321 UK locations on three delivery platforms, automated monitoring brought stores back online 892 times in a single week, and averages about 4,000 restored connections a month. Uptime moved to consistently above 99% within six months.

Jon Longobardi, Head of Digital Operations at Burger King UK, put the old situation like this: "Downtime used to be a silent killer." His complaint about the reporting is the one every multi-unit operator will recognise. By the time the number showed up in a dashboard, the money was already gone and there was nothing to do with the information except feel bad about it.

A setup that works is boring and cheap. Every platform's merchant app on a phone belonging to someone with authority, notifications on, each platform tracked separately. An integration check written into the opening routine and repeated mid-shift. Store hours audited against reality on every platform, including the one you set up two years ago and never opened again. Alerts going to a person who is in the building, not to a regional inbox that gets read on Monday.

Then fix the causes in the order they happen, not the order they feel urgent.

Hours come first, because they start the chain. Wrong hours produce driver reports, driver reports produce cancellations, cancellations produce deactivation. DoorDash applies an automatic buffer to stop orders landing right before close, but only against the hours you gave it. Holidays and early closes for a private event have to go in before the day, not during it.

Out-of-stock discipline comes second. An item left marked available after it ran out creates an avoidable cancellation, and two of those in a day can take the store down. The person who knows the shrimp ran out at 7:15 is a line cook with eleven tickets in front of him. Whether that fact reaches the tablet is a training problem, not a software problem, and no amount of middleware fixes it.

Pause hygiene comes third. Teach people when to pause, then build a rule that nothing stays paused past the end of a shift without a manager clearing it. Most forgotten pauses are not carelessness. They are a 10 p.m. handover where the person who paused the store went home and the person who arrived never knew.

Integration monitoring comes fourth, and it is last only because it is rarest. Weak Wi-Fi at the pass, expired credentials, a POS that quietly stopped acknowledging orders. From the floor all three look the same. They look like quiet.

A number the client has never seen.

Ask a restaurant owner how many hours they were invisible last month. Almost none of them can tell you. The data is sitting in the merchant portal: DoorDash shows downtime and avoidable cancellations under Operations Quality, with the lost-sales value of cancelled orders attached to it. Hardly anybody opens that tab. Pull the report during a sales call and you stop talking about features.

The math closes itself at small scale. A ten-unit group at the poor-performer average is looking at $170,000 a year. Monitoring that recovers a third of it pays for itself several times over, and the operator sees the difference in the first month rather than the second year.

There is a channel argument too. Payment residuals keep getting squeezed, and uptime monitoring is recurring revenue that nobody is fighting you over yet. You are not selling software to a restaurant. You are selling back the hours it was open and nobody could find it.