The Reason Americans Have Stopped Drinking Out

And what it signals for the restaurant trade elsewhere

At every single conference or wine fair I’ve attended this year, someone will raise an alarm about GLP-1 drugs. Medications like Ozempic reduce the desire for alcohol and so, the argument goes, when enough people are taking them, drinks sales will crater.

It came up again while I was moderating the panel on fine wine at the Agora forum at ProWein. Could it be that these drugs are already having an impact, and that’s why sales are falling?

It’s a reasonable fear. The industry is already anxious about declining consumption, so a widespread drug that reduces the desire to drink is a nightmare scenario.

Bank analysts tell a different story. With some caveats, they point to the cost of living as the main driver of falling sales, particularly in the on-trade. Then again, they’re not the ones offering wines to customers who don’t seem interested.

On my train home from ProWein, I read a New York Times piece on exactly this question — and found the readers’ comments more illuminating than the article itself.

The big question

The NYT story is called Dinner and No Drinks: Restaurants Are Struggling as Americans Drink Less. The author, Meghan McCarron, interviewed restaurateurs, wine directors and others involved in the on-trade, to get a sense of what was happening.

“According to David Henkes, a senior principal at the food service research firm Technomic, alcohol sales are slumping in every category of the restaurant industry, from fine dining to casual establishments, with 31 percent of operators reporting ‘severe declines’ in alcohol sales in 2025,” she wrote.

Falling alcohol sales are a catastrophe for the restaurant world, because alcohol subsidises everything else, given that rent, labour and ingredients cost too much to generate much profit, if any.

Both the author and those she interviewed speculated about the reasons customers were drinking less, from the rise of GLP-1s to new moderation behaviours, to Gen Z not touching the stuff, to the cost of living.

So what’s the answer?

Asking the people who are living it

Personally, I would plump for price being the worst offender. I’ve been to the US three times in the last six months, and every time I’ve been shellshocked by what it costs to enjoy even a modest dinner. The wine is sometimes the worst value thing on the menu, unfortunately.

But that’s just my experience as a jet-lagged foreigner. How do locals explain what’s going on?

NYT readers sure have strong opinions on the subject. More than 1,100 of them waded in to offer an opinion, enough of a sample size to draw some conclusions.

Obviously it’s a self-selected group and there’s no control, but it’s also true that these are committed middle- and upper-middle class restaurant-goers, of the type the on-trade can’t afford to lose. That makes them worth listening to.

The verdict from the comments

The answer, decisively, was cost. About 75% of the comments, some of which had real anger in them, focused on price:

“$15-20 for a shot of mass-produced Scotch.”

“$25 cocktails and $90 for a meh bottle.”

“Whole bottle of same wine goes for $18-20 in the liquor store.”

“I am not paying $20 for a mediocre glass of wine.”

It wasn’t just the price of the drinks. Dozens of commenters talked about how tipping culture was pushing the cost of food and drink into unmanageable territory.

About 18-20% of the commenters said they either now stayed home altogether, or drank before they went out to dinner; some complained that servers then treated them badly when they didn’t order alcohol.

There were a few comments about health, including people who said they could no longer drink because of age or medication, accounting for around 18% of the total. Some also said they were substituting alcohol with cannabis, but this was a small number.

And, of course, there were some anti-alcohol comments, plus some comments from scolding types who couldn’t miss an opportunity to vent their disapproval:

“Fewer restaurants would mean healthier Americans, less diabetes and less heart disease.”

But these were a minority. The majority talked about unsustainable costs.

Admittedly there’s a structural factor the comments couldn’t capture: the NYT piece notes that Millennials are now drinking less and haven’t been replaced by Gen Z, which is a demographic story that will unfold regardless of what restaurants charge.

But my overall take is that while health, wellness and medication are clearly playing a role, they’re not the main story. It’s not the GLP-1s, or GenZ or cannabis that’s killing the on-trade, but the prices.

I’m pretty confident about that conclusion. Getting there, however, wasn’t easy.

Ten minutes, give or take seven hours

As my train home from ProWein in Düsseldorf had WiFi, I decided to stick all the comments into AI, get it to analyse them, and then give me a breakdown of the results, thinking this project would take about ten minutes.

It took me an entire day.

First, the WiFi wasn’t strong and every time I scrolled to the end of the comments, it would cut out just before I could copy and paste. Then it turned out I couldn’t just drop 73,000 words into ChatGPT and get it to arrange everything into a nice chart.

ChatGPT actually got a bit snippy and told me I needed to load everything into Excel before it would complete the task, and produced a long doom scroll of instructions for me to follow. (Which I didn’t.)

NotebookLM, famous for its ability to deal with big files, got stuck on a commenter from Philadelphia who had tallied up 400 comments, and just kept serving me up his analysis — which later turned out to match what the overall data showed, more or less exactly.

Finally, Claude gave me some useful instructions on how to break up the comments and get what I wanted.

Then I checked everything with ChatGPT, because despite the computing power of these models, none of them can be trusted with basic addition. I ended up doing a lot of counting by hand.

(I have this conspiracy theory that the LLMs are trained to undermine each other, because ChatGPT was scathing about how Claude.ai segmented things.)

In the end, both LLMs analysed an average of ~840 comments each.

A non-American in Düsseldorf

A number of commenters also compared American prices to the much more affordable wine prices they experience at restaurants in Europe.

Reading them made me reflect on my experience during ProWein, when I stayed at an American hotel chain. I went downstairs to the bar on Saturday to catch up with a colleague, and bought us a very mediocre glass of Riesling for €15 each. That price would be unremarkable in Manhattan, but is almost double the local price. I am guessing it was priced like that because the hotel caters to international business travellers used to American prices.

Since Covid, American payment technology has also spread across Europe, and with it comes the tip-prompt screen, which presents 20%, 25%, and 30% as the default options. Now you stand there feeling exposed while you stab the thing trying to bypass these options, none of which represent local norms. These machines are not just importing a payment system, but also an assumption about how pricing should function.

I’ve also noticed that taxi drivers and waiters now pointedly ask me for tips once they hear my non-German accent and assume I’m American; Germany has traditionally either not expected tips, or kept them at under 10%. I noticed this at Wine Paris as well, where taxi drivers going past in the evening would announce an outrageous price up front, many times what the metered charge would have been.

The real question for the on-trade, then, is whether what’s happening is a uniquely American problem or an early warning. Restaurants here in Europe are also suffering from higher ingredient and labour costs, and I worry that some of the same economic forces that are driving Americans out of restaurants are embedded in this ubiquitous hardware. The European industry might be watching its own future play out across the Atlantic.

Have You Taken a GLP-1? Or Know Somebody Who Does?

There’s a lot of anxiety about GLP-1 medications (Ozempic, Wegovy, Mounjaro etc) and their impact on wine, spirits and beer consumption. But the existing research tends to treat GLP-1s as a single mechanism producing a single effect, when the reality is much more varied.

If you’ve ever taken a GLP-1 medication (even if you’ve stopped!) I’d like to hear what it did or didn’t do to your drinking. Both answers are useful.

The survey takes around five minutes and is completely anonymous. I have no way of seeing your name or email address.

https://forms.gle/y8u3SzeGc7uV3R2j9

The results will form the basis of a Wine Business Monthly story, as well as future editions of Drinks Insider.