Three Restaurant Statistics Everyone Quotes. We Couldn't Find the Study Behind Any of Them.

Start with the one you have definitely seen.

"Adding photographs to your menu increases sales by 30 per cent." It appears in hospitality trade articles, in software marketing, in consultancy decks, and in a great many LinkedIn posts. It is usually stated without a source. When it does carry one, the source is another article stating it without a source.

We followed the trail as far as it goes. There is no study. Every instance resolves into a marketing article citing a marketing article citing a marketing article. Somewhere at the bottom, someone made the number up, and the industry has been passing it around ever since.

The same is true of "QR menus increase average order value by 28 per cent." We searched specifically for peer-reviewed research measuring the effect of QR menus on order value, order accuracy, table turnover or labour efficiency. There is none. Not a weak study — no study. It is a genuine hole in the literature.

And "printed menus cost the average restaurant around $1,200 a year." Every figure of this kind in circulation traces to a company selling digital menus. We checked a commercial menu printer — a business with precisely the opposite incentive — and its published cost page contains no per-unit pricing at all.

We publish digital menu software. Two of those three numbers exist to sell what we sell. We still can't use them, because they aren't true.

What is actually there, and why it is better

Menu design has an unusually rich body of genuine experimental research behind it. It is just not the body of research being quoted.

Price format: about 8 per cent, measured in a real restaurant

Cornell researchers ran a field study with 201 diners comparing three price formats: numerals with a currency symbol, numerals alone, and prices written out in words. Diners given numeral-only prices spent approximately 8 per cent more — an average of $5.55 more per person.

The result was counterintuitive in a way that makes it credible. There was no difference between the currency-symbol format and the written-out format. The effect appeared only when the symbol was removed entirely. Sheryl Kimes's explanation: "references to dollars, in words or symbol, reminds people of the 'pain of paying'."

Eight per cent, from deleting a character. That is a smaller number than "30 per cent" and an infinitely more valuable one, because it is real.

Position: up to twice as popular, replicated in the field

Dayan and Bar-Hillel, publishing in Judgment and Decision Making, found menu items placed at the beginning or the end of their category list were "up to twice as popular" as the same items placed in the middle.

The measured swing was ten percentage points — 45 per cent when centred, 55 per cent at the edges — and it replicated in both a laboratory study of 240 participants and a field study of 459 orders on a baseline menu against 492 on a manipulated one, across thirty study days in a working café.

And the one that should have changed the industry

For decades, menu design convention has held that there is a visual "sweet spot" — above the midline, right-hand page — where diners' eyes settle, and that this is where high-margin items belong. Consultancies have been charging for this advice since before most of us were working.

An infrared eye-tracking study found no evidence it exists. Diners read menus sequentially, like a book, left to right and down the page. The study did identify a low-attention "sour spot" — which, on the menu tested, contained the restaurant's information and a list of salads.

This is the most commercially significant finding in the menu-design literature and the one least reflected in practice. Sequential reading means order governs attention — which is exactly the variable Dayan and Bar-Hillel measured, and exactly the variable a digital menu can change without a reprint.

So what about photographs?

There is real research here. It is conditional, which is presumably why the invented version travelled further.

Hou, Yang and Sun, in the International Journal of Hospitality Management, tested whether pictures help. Their answer depends on the dish's name.

For dishes with common, descriptive names, adding a picture improved attitudes towards the item, willingness to pay, and purchase intention. For ambiguously named dishes, pictures helped only some consumers — and visual processors showed less favourable outcomes with pictures than without.

So the implication is not "photograph everything." It is: photograph dishes whose names already describe them, and fix the name before you fix the photograph. A picture attached to an opaque or invented dish name may perform worse than no picture at all.

That is a harder rule to put on a sales slide. It is also actionable in a way that "30 per cent" never was.

Why any of this matters: 2.8 per cent

The National Restaurant Association's operations data, drawn from more than 900 restaurants, puts median income before taxes at 2.8 per cent of sales for full-service operators and 4.0 per cent for limited-service.

Read that against the same dataset's cost lines. Food and non-alcohol beverage costs run a median 32.0 per cent of sales. Salaries and wages run 36.5 per cent.

At a 2.8 per cent margin, one percentage point of movement on the food cost ratio consumes roughly 36 per cent of the entire pre-tax margin.

That is the reason menu decisions are financial decisions rather than design decisions. It is also the reason the interesting question about a printed menu is not what it costs to print.

The real cost of a printed menu isn't printing

In August 2026 the FAO Food Price Index averaged 133.3 points — up 1.9 per cent on July, a stable enough headline. Underneath it, the sugar sub-index rose 11.9 per cent in a single month.

US data tells a similar story at the till: the Bureau of Labor Statistics recorded full-service meal prices up 3.5 per cent over the twelve months to August 2026, and the USDA forecasts food-away-from-home prices up 3.6 per cent across 2026.

Now consider what a printed menu can do about that. On a twice-yearly reprint cycle, the average unrecovered lag on any affected item is three months. Input costs move monthly. The page does not.

We call this repricing latency, and it is the honest version of the argument the invented $1,200 figure was standing in for. We cannot tell you what your menus cost to print. We can tell you that a static page in a business with a 2.8 per cent margin carries a structural delay between a cost moving and a price responding — and that the arithmetic on that delay is not in dispute.

One study we deliberately left out

You will find, in almost every article on menu design, the finding that descriptive menu labels increase sales of the labelled items by 27 per cent. It is a 2001 paper, and it is a good story.

We excluded it. The paper itself carried no retraction notice when we checked, but its lead author was subsequently found by his own university to have committed academic misconduct including "misreporting of research data" and "problematic statistical techniques."

We judged the figure unusable. You will encounter it everywhere else; you will not find it in our report.

And one gap we are admitting

We publish QR menu software. There is no peer-reviewed study measuring what QR menus do to average order value, order accuracy, table turnover or labour efficiency. We looked hard, because we would have liked to find one.

The nearest rigorous evidence covers digital ordering modes generally. Research in the Journal of the Academy of Marketing Science, reviewing more than 23,000 orders across six field studies, found that digital ordering "leads consumers to have a more automatic decision making mode and lower cognitive involvement, which results in more indulgent outcomes in the form of unhealthy food orders and higher overall spending." The authors publish a magnitude for the dietary effect. They publish none for the spending effect.

And there is research pointing the other way: a 2024 study in the Journal of Hospitality and Tourism Management found that QR code menus diminish customer loyalty relative to traditional menus, through perceived inconvenience, moderated by the customer's need for interaction.

That finding is in our report too, with a section of its own. It is the single most important thing a restaurant considering a QR menu should read, and the mechanism it identifies — friction, concentrated among guests who want human contact — is something you can design around rather than something that ends the conversation.

The case for digital menu infrastructure does not need invented statistics. We made it without them, and it came out stronger.