A call that doesn’t become an order still tells you something.
Today that something disappears the moment the caller hangs up — not because anyone did anything wrong on the call, but because nothing exists to hold onto it once the conversation ends.
A complete, ordinary call,
handled honestly and well — and gone.
A caller asks whether the restaurant does a specific vegan option — something not on the menu tonight, maybe never has been. Whoever picks up says no, apologizes, and the call moves on to what the caller orders instead, or ends there if they don’t order anything at all. That is also the entire record of what just happened: nobody wrote down that a vegan option came up, because there was no order to attach it to and no form built for a request that came to nothing.
If the same question comes up nine more times that month, from nine different callers, nothing connects any of them to any of the others. Each lands, gets answered honestly, and disappears the moment the call ends — unless a person happens to remember it, and happens to mention it to someone who could act on it.
The gap is not that anyone handled a caller badly. It is that a restaurant’s only durable record of its own demand is the demand that already turned into a sale. There is no shared place a manager could look later to see what has been asked for and not offered, and no way to tell, three weeks on, whether last Tuesday’s question was a one-off or the fourth time this month. Everything else lives, at best, in one person’s memory of one shift.
More than a quarter of calls.
Plausibly approaching a third.
Two things push the honest range higher than that second figure on its own would suggest. First, “revenue-tied” is a wider net than “an order” — a reservation call counts inside the 71% even though it never becomes something a kitchen fills, so the true non-order share is larger than 29% implies, not smaller. Second, the restaurants in that sample lean full-service, where reservations make up a real share of what counts as revenue-tied in the first place. For a quick-service counter or a pizza shop, where almost nobody calls to reserve a table, the same figure would plausibly run higher again.
Put together, the defensible range is more than a quarter of calls, plausibly approaching a third or more. Nothing in the available research supports rounding it any higher, however tidy a round fraction might look on a page. Every one of those calls is a real conversation that happened; this page is about what is currently done with the ones that don’t end in a sale, which today is nothing at all.
Most of what a restaurant sells
no longer happens where an owner can watch it.
However that share actually breaks down for a given restaurant, it means a majority of what a restaurant sells doesn’t happen where an owner can simply look up and watch it happen — which is exactly why what a call reveals matters as much as what it eventually sells.
Three ordinary decisions,
all made a little blind.
A menu decision gets made with less than it could have when the requests that never became orders aren’t part of what informs it. A staffing decision gets made the same way — nobody weighing a real signal, just a hunch. So does a purchasing call: whether to stock an ingredient, or list an option that keeps getting asked about anyway. None of that has an honest dollar figure attached here. Nobody has measured what a restaurant loses by not knowing what it doesn’t know, and this page won’t invent one to make the point land harder.
What changes is narrow, on purpose: nothing about how a call gets answered is any different. What is different is what happens to a request after the call ends. When a caller asks for something that isn’t on the menu, or asks whether something is available, that moment is kept as its own fact — tied back to the call it came from, held in the caller’s own terms rather than folded into a vague sense that people ask for odd things sometimes.
The different kinds of question a phone line actually gets — hours, an address, a price, delivery area and fee, a promotion, an order status check, a request for a manager, a vendor call — are recorded as what they actually were, not lumped into one undifferentiated bucket. And a restaurant can ask something of that record rather than being handed one fixed report. The full shape of how asked and missing is built, and exactly what it can and can’t tell an owner, belongs on intelligence — this page is about the situation it answers, not a second explanation of the mechanism.
A checkable list.
Not a score standing in for one.
An owner keeps hearing, secondhand, that people sometimes ask whether delivery reaches a specific part of town — and wants to know whether that is actually true, or a story that stuck around without much behind it. They pick the neighbourhood by name and set the range to the last two months.
What comes back isn’t a percentage or a score standing in for the real thing. It is a specific, checkable list — every call in that window where a caller asked about delivering to that area, each one linked back to the call it came from. Maybe three. Maybe eleven. Either way the owner has something to act on that isn’t a guess and isn’t a rumour.
What they don’t get is a prediction of next month’s demand, a ranking against how other restaurants perform, or a single headline number standing in for all of it — because none of those exist to show honestly, and this page doesn’t invent one to fill the space.
Objections, answered
Doesn't this just predict what's coming?
No. Everything kept here describes a call that already happened. Nothing forecasts next week, next month, or a busy Friday before it arrives.
Isn't that the same as reading through call transcripts?
Not quite. A transcript is one call, read on its own, one at a time. This is many calls at once, filtered by what was actually asked for, rather than read one conversation after another.
What if a caller's request was too garbled to make out?
It isn’t silently counted as a specific miss. Only a request the system actually determined wasn’t on the menu gets kept that way — a genuinely unclear moment isn’t dressed up as a clean data point it never was.
Does noticing a pattern change the menu by itself?
No. This only reports what already happened on real calls. Adding an item back, or adding a new one, is still a decision a person makes on the menu itself.
Nowhere here, and nowhere on the page this hands off to, is there a figure, a percentage, a ranking, or a comparison describing what a restaurant’s own kept requests actually add up to — not an average, not a most-requested list with numbers attached, not a comparison to any other restaurant’s line. What is real is the capability itself: a request gets kept, tied to its call, and can be asked about on the restaurant’s own terms. What isn’t claimed anywhere is a demonstration of what it would show for any given restaurant, because that depends entirely on that restaurant’s own calls.
RELATED
Intelligence
Every call becomes data you own: demand, asked-and-missing, the closeout ledger.
OPEN →PLATE Nº 009Questions
Hours, parking, gluten-free — answered from what you actually published, never guessed.
OPEN →PLATE Nº 030Missed calls
Every ring answered — even the ones that used to hit voicemail.
OPEN →Find out what your own line has already been asked.
Request access and start keeping what your calls reveal, not just what they sell.