What guests almost order: the signal your till never sees

Your till is an excellent historian. It tells you, to the cent, what happened last night: forty covers, nineteen plates of the ragù, four of the venison. What it cannot tell you is what almost happened — the dish a table looked at, liked enough to shortlist, and then quietly dropped before calling the waiter over. That moment leaves no trace anywhere in your business. No order, no receipt, no line in the report. It simply evaporates.

A digital menu is the one place where it doesn't have to.

Where the signal comes from

A guest scans the QR code on the table and opens your menu. As they read, they can tap a heart on anything that appeals — building a little shortlist to show the waiter when they order, instead of scrolling back up and going "sorry, what was the one with the fennel?".

That's it. That's the whole mechanic. No account, no login, no personal data: nobody is identified, nothing is stored about the person. What the menu keeps is a count per dish, and nothing else.

Three gestures, and only three

Menudetto counts exactly three things:

• Added — the dish went into a shortlist.
• Removed — it was taken out again, before the list was finished.
• Confirmed — the guest asked for the final list, ready to order.

Added and removed, together, are the interesting pair. That combination is a second thought, and it is the part of the evening no other system in your restaurant records. A till, a POS, a management suite — they all register what happened. They cannot register what was set aside. There is no field for the dish somebody wanted for thirty seconds.

Say it plainly: these are not sales

This deserves its own heading, because it is the single easiest thing to get wrong.

These numbers are not orders, and they are not revenue. Nothing arrives here from your till. Menudetto has no idea what was actually served or what was actually paid. A dish with a lot of hearts has not sold more; a dish with none has not sold less. The page itself says so, in as many words, so nobody reading it over your shoulder gets the wrong idea.

Think of it as a second, separate instrument. Your till measures results. This measures intentions before the result — and the two answer different questions. Reading the hearts as if they were a sales chart throws away the only thing that makes them worth having.

Below twenty lists, the page refuses to rank anything

Small numbers lie confidently. Three people tapping a heart on the octopus tells you nothing at all, but a bar chart drawn on top of those three taps looks exactly as authoritative as one drawn on top of three hundred.

So under twenty shortlists prepared at the table, the page will not produce a ranking. It shows you how many are still missing and stops there. As it puts it: three hearts are not a trend.

That refusal is worth more to you than a chart would be. A ranking built on a handful of taps would have you rewriting a dish that nobody has really considered yet — spending real kitchen and menu attention on noise. Waiting a couple of weeks costs you nothing; acting on nothing costs you a week of work in the wrong direction.

Above the threshold: four groups, one piece of advice each

Once there's enough to say something, every dish lands in one of four groups.

They think about it, then take it off

The dish gets shortlisted regularly and then removed before the order. People are interested — the idea works — and something at the last moment doesn't. Almost always that something is how it's written or what it costs. The advice is to rewrite it: say what's in it, where it comes from, why it's worth the price. Our guide to writing dish descriptions covers the wording; the price is your call, but this is the group where a price sitting slightly wrong shows up first.

It never gets into the lists

Hardly anyone shortlists it at all. Here the problem is usually not the words — nobody read them. It's position. The dish is at the bottom of a long section, or in a section people reach after they've already decided. Move it up: open a section with it, or lift it above the items that are already doing fine without help.

They choose it and confirm it

Shortlisted, kept, confirmed. Leave these alone. No new wording, no repositioning, no clever experiment. The most common way to damage a menu is to improve the part of it that was already working.

The rest of the menu

Everything else. No advice at all, on purpose: the signal isn't strong enough to say anything honest about these dishes yet. An empty recommendation is better than a confident wrong one.

Reading a single dish

Each dish gets one bar. Its length is the number of shortlists that took the dish in; the red section at the front is the share that later dropped it. Underneath, the raw counts in plain figures — added, removed, confirmed — so you're never reading a shape without the numbers behind it. And again, on the page itself: those are lists, not covers.

Where menu engineering comes in

This is the complement to a method you may already use. Menu engineering crosses sales against margin: what earns, what shifts, what to drop. It works on completed transactions — the historian's view.

Second thoughts sit upstream of that. A dish your margin analysis calls a puzzle — good margin, barely sells — reads very differently depending on which group it lands in here. If guests shortlist it and then remove it, the dish is being seen and rejected: rewrite it. If it never enters a list at all, it's being missed entirely: move it. Same symptom in the accounts, two completely different fixes, and only one of the two instruments can tell them apart.

One requirement

The menu has to be published, with the QR codes actually on the tables. There is nothing to count until people are reading it in the dining room — this is a live-service signal, not something you can generate from a spreadsheet.

In short

Your till records what happened. The shortlist records what was set aside, and nothing else in your restaurant does. Not sales, not orders, not revenue — a different measurement, of the moment just before the decision. Below twenty lists the page keeps quiet; above it, you get four groups and three sensible moves: rewrite the dish people consider and drop, lift the dish nobody sees, and don't touch the ones that are working.

If you want to see it on your own menu, create yours free, publish it, and put the QR codes out. The counting starts with the first table.