85% of People Answer a Call From a Courier Who Isn't Real

We run a food delivery site that never delivers any food. People browse dishes from 18 cuisines, fill a cart, check out with a card that isn't charged, and watch a courier cross a map towards an address that doesn't receive anything. We have now been doing this long enough to have real numbers about it, and some of them surprised us.

What follows covers 26 June to 29 August 2026: 209,316 people and 103,648 imaginary orders. These are our own analytics, aggregated, with no personal data involved. We have also been honest at the bottom about the things this data cannot tell you, because that turns out to be the more interesting half.

Nearly everyone answers the phone

Partway through the delivery, the courier calls you. It is a recording. Nothing is at stake. You know, with complete certainty, that there is no driver, no food, and no reason to pick up.

The phone rang 14,173 times. It was answered 12,031 times.

We did not expect that. If the fake order were purely a joke, the rational move is to let it ring — you already know the punchline. Answering means something else is going on: for the length of that call, people are choosing to stay inside the fiction rather than break it. The ritual is doing more work than the gag.

But most people don't wait for the food

Of the 103,648 orders placed, only 37,832 were still being watched when the courier arrived. That is 36.5%. Almost two thirds of people place an order and leave before the delivery completes.

Read those two findings together and the shape of the thing becomes clear. The satisfying part is not receiving. It is ordering — the browsing, the cart, the checkout, the small ceremony of having decided. Once that is done, the craving is largely spent, and the arrival is optional. The one moment that reliably pulls people back is the phone ringing.

Almost half of the orders end in an actual recipe

This is the number we care most about. Alongside the fake order, every dish links to a real, free recipe. Over the same period there were 49,274 recipe opens against 103,648 orders — roughly one recipe opened for every two fake orders placed.

We built the recipe collection as an exit ramp: if the craving turns out to be genuine hunger rather than the itch to browse, there should be something on the other side of it besides a delivery fee. We assumed it would be a rounding error next to the novelty. It is not close to a rounding error.

Twice as many people turn notifications off as on

We can send a notification when the imaginary food is on its way. 3,463 people opted in. 6,727 opted out — about two refusals for every acceptance.

People will happily spend ten minutes inside the fiction, and answer the courier's call while they are in it, but they will not let it follow them out of the tab. That seems like a healthy instinct, and it is a useful reminder for anyone building this kind of thing: the appetite for the ritual is not an appetite for being pestered.

The features nobody asked for

Party mode lets several people order together. In two months it was started 442 times and joined 375 times, against 103,648 solo orders. That is about 0.4%.

Ordering imaginary food, it turns out, is something people do alone. We are not deleting it, but we are not going to pretend it found an audience either.

The "order me anything" button — which picks a dish for you — was started 6,413 times and carried through to a pick 2,650 times: 41%. The majority of people who ask to be surprised change their mind and go back to choosing for themselves. Even here, with nothing on the line, people want to pick their own dinner.

Where in the world

Orders came from everywhere, led by the United States (27.8%), then the United Kingdom, Malaysia, China, Indonesia, Brazil and Vietnam. Seeing Malaysia and Indonesia that high, ahead of much larger markets, was not something we predicted.

76.4% of orders were placed on a phone. This is a thing people do standing in a kitchen at the wrong hour, not sitting at a desk.

What this data cannot tell you

We would rather say this out loud than let anyone quote a number we do not stand behind.

• We cannot tell you what time of night people order. Our analytics record the hour in a single fixed timezone, so a global audience gets flattened into one clock — the apparent peaks are mostly just different continents waking up. The obvious headline here would have been about 2am cravings. We do not have the data to write it.
• We cannot tell you which dish is most ordered. The dish name was being collected but not in a form we could query, and that is not retroactive. We have fixed it going forward, which means this question gets an answer later this year and not before.
• One day is an outlier. 16 August produced about 25 times a normal day's orders. We have left it in the totals because it happened, but we have not built any claim on the daily or weekday patterns, which it badly distorts.
• Some counts start later than others. Not every event has been recorded since the first day the site collected data, so treat comparisons between two different events as approximate rather than precise.

The short version

People come for a joke, stay for the ritual, answer the phone to a driver who does not exist, leave before the food they were never going to get arrives, and then — about half as often as they order — go and look up how to actually cook it.

That last part was the hope when we built this. It is quietly the most encouraging number on the page.

You can go place an order yourself. It is free, nothing is charged, and no food will come.

Quick answers

How many people use FoodNeverComes?
Between 26 June and 29 August 2026, 209,316 people placed 103,648 imaginary orders on the site.

Do people really answer the fake courier's call?
Yes. The in-app courier call rang 14,173 times and was answered 12,031 times — about 85% — even though everyone knows the courier and the food are not real.

Does anyone actually cook the food afterwards?
Every dish links to a free recipe, and there were 49,274 recipe opens against 103,648 fake orders over the same period — roughly one recipe opened for every two orders placed.