Everyone in my trade knows what the weight loss drugs are doing to restaurants. Smaller appetites, smaller checks, empty seats at nine o'clock. I have heard it in three kitchens this year, always with the same shrug, the shrug of a man describing weather.
Then in May the National Restaurant Association published a survey, and the number in it was the opposite of the thing everyone knows.
People taking a GLP-1 bought a meal, a snack or a drink from a restaurant seven point six times in the previous week. People not taking one did it five point one times.
They are not staying home. They are coming half as often again as the rest of us.
I want to take that number seriously, and then I want to take it apart, because it is the most flattering number my industry has been handed in years and we have been repeating it without asking who wrote it down.
The part that is probably true
Roughly one in eight American adults is on one of these drugs now. That is not a niche. The same survey found eighty seven percent of them say they enjoy going to restaurants, and seventy one percent call restaurants an essential part of their lifestyle, against sixty one percent of all adults.
And the spending tells a second story that the visit count does not. Separate research puts GLP-1 users at about twenty one percent fewer calories and roughly a third less money spent on food.
Put those together and you get something specific. Not fewer customers. More visits, each one worth less.
You can see it from the host stand. The four top that shares three starters, orders one glass each and a large sparkling water, and stays two hours. Nobody at that table is unhappy. Everybody at that table is spending less than they used to.
Anyone who has run a dining room can feel what that does. Your covers hold or rise, which is the number on the wall that everyone watches. Your average check falls, which is the number that pays for the building. You are busier and poorer at the same time, and the busyness disguises the poverty for about two quarters.
It is the wine story, running backwards
I wrote recently about what happened to drinking. Participation held. Frequency collapsed. Daily drinking among the young went from six percent to two, and the thing that died was the habit, not the category. Fewer occasions, each asked to carry more weight.
The restaurant version is that film played in reverse. Drinking went from habit to occasion. Eating out, for this group, is going from occasion back to habit. More often, less each time, less ceremony, less spend. The seven point six is a habit number. Nobody has a special night out seven point six times a week.
Which is the part that should worry an operator, because the whole of fine dining is built on the occasion. The wine pairing, the tasting menu, the two hour table, the reason the room exists in the form it does. A habit customer is a good customer and they are not that customer, and a business designed for one does not quietly convert to the other.
Now look at who published it
The survey is by the National Restaurant Association. The headline on it is *"Shifting Habits, Not Shrinking Demand."*
That is an industry body writing about its own industry in the middle of a scare. Nobody falsified anything. The number is real and the framing is interested, and my own trade did the same thing with wine for a decade.
Here is the specific problem.
Seventy one percent of GLP-1 users call restaurants essential to their lifestyle. Sixty one percent of all adults do.
That gap is presented as a finding. It is at least as likely to be a description of who these people were before they filled a prescription. This is a drug that costs real money every month, taken disproportionately by people with insurance, disposable income and an appetite for managing themselves. Ask whether those people eat out more than the general population and the answer was always going to be yes.
The survey may have discovered that restaurant people take GLP-1s. It is being read as evidence that GLP-1s make people restaurant people. Those are different sentences and only one of them is in the data.
I cannot prove which it is, and neither can the survey, because it compares users to non-users rather than to themselves before they started. That is the study nobody has run, and until somebody does, seven point six is a fact about a population, not an effect of a drug.
The drinking side
The alcohol evidence is better and the headline is worse. In May the Lancet published the largest trial yet: a hundred and eight adults in Copenhagen with obesity and alcohol use disorder, half on semaglutide, half on placebo, everyone getting talk therapy, for twenty six weeks.
Heavy drinking days fell by forty one percentage points on the drug. They fell by twenty six point four on the placebo.
Only the first number travelled. The second is the quieter finding: people who got a placebo and some structured conversation cut their heavy drinking by more than a quarter. The drug is a real force on drinks volumes. People deciding to drink less, with support, is the bigger one, and it was true before any of this existed.
What it is actually for
I wrote once about a woman putting on a costume because the self underneath had been publicly renamed, and how the interesting part was never the provocation but the choreography. The role as shelter. The performance as a way to be legible again.
These drugs are that move, made chemically and made available to everyone. They are the most efficient self-optimization tool ever sold, and self-optimization has been the organising project of this decade the way home ownership was for another one. You are not fixing a medical problem, or not only. You are stepping into a version of yourself that the culture currently reads as competent.
And a role needs somewhere to be performed. You cannot do it alone in a kitchen.
That, I think, is what seven point six is. Not appetite. Attendance. The restaurant is where the new version gets seen, ordering carefully, eating a third of it, staying an hour. The visit is the point and the food is the evidence.
It explains the rest of the survey too, which otherwise reads oddly. Eighty nine percent of these guests say restaurants could do a better job highlighting the healthy choices. Seventy six percent say they would pay more for something customised to them. They are not asking for less food. They are asking to be recognised.
What I would do about it on Monday
Three things, and none of them are a menu.
Stop watching covers. Covers will hold and they will tell you nothing. Watch average check, and watch it by daypart, because the habit visits and the occasion visits are now different businesses inside one room and the blended number hides both.
Build something worth a smaller appetite. Not a diet menu, which is a way of telling a guest you have noticed their body. A real half portion at a real price, and a by the glass list deep enough that one careful glass is a considered choice rather than a consolation. The guest ordering one thing well is the guest with the highest margin in the room and most lists are built to punish them.
And do not let the covers number get quoted at you. It is the one the trade body is holding up, and it is the one that stays healthy longest while the business underneath it changes shape.
The habit I would rather we kept
I do not say any of this to be superior about a survey. Every operator I know, me included, has repeated a comforting number without asking who chose the question.
The discipline is small and it is not hard. When a number arrives that tells you the thing you were afraid of is not happening, find out who published it before you repeat it. Not because they are lying. Because the question they chose to ask is the whole answer, and a trade body asking whether demand is shrinking will not ask whether it is changing into something less valuable.
Demand is not shrinking. That much I believe.
It is getting more frequent, and thinner, and it is walking in the door on a Tuesday wearing a version of itself it paid for, and it would like one glass and the good half portion and to be seen having a nice time.
We should probably build the restaurant for that, rather than reassuring each other with the attendance figures.