The Ozempic effect on food: what GLP-1 medicines are doing to groceries, restaurants and packaged food

Three claims inside one nickname

GLP-1 medicines such as semaglutide (Ozempic, Wegovy) and tirzepatide (Mounjaro, Zepbound) turn down appetite. The knock-on effect on what the country buys has picked up a nickname, the Ozempic effect.

That nickname bundles three different claims. The first is that households buy less food. The second is that restaurants and manufacturers are changing what they sell. The third is that a new category of "GLP-1 friendly" products exists. The first claim has real purchase data behind it. The second shows up in menus and earnings calls. The third is mostly a labeling question, and the evidence there is thinner than the marketing suggests.

Atlas does not sell or recommend products. Where a company makes a claim, the page says it is the company's claim. For the medicines themselves, see the Business and markets hub and the guide to what to eat on a GLP-1 medicine.

How many people are we talking about?

Every number in this story depends on how many shoppers are on the medicines, and the published counts differ because they count different things.

• KFF poll: US adults taking one, any reason (adults)12%
• Gallup: US adults taking one for weight loss (adults)11%
• Circana: US households with at least one user (households)23%
• PwC via Reuters: US households with at least one user (households)about 20%
• Worldpanel: Great Britain households with at least one user (households)6.3%

Bar length is the share reported, on a 0 to 25% scale. The bar color changes with the unit, adults or households. The first four rows are the United States, the last is Great Britain.

A household count runs higher than a person count, because one user puts a whole household in the tally. A company that says a fifth of homes now include a user is describing something different from a fifth of shoppers cutting back. Surveys also differ on the question: KFF asks about any reason, Gallup about weight loss. Adoption is still rising, so older figures understate today's.

What purchase data shows about groceries

The strongest US evidence so far is a study by Sylvia Hristakeva and Jura Liaukonytė at Cornell's Dyson School, with Leo Feler of Numerator,, published December 18, 2025 in the Journal of Marketing Research. They matched purchase records from Numerator's panel of about 150,000 households with repeated surveys asking whether anyone in the household took a GLP-1 medicine, and when they started.

• Grocery spending, all households−5.3%
• Grocery spending, higher-income households (more than)−8%+
• Limited-service restaurants, meaning fast food and coffee shops (about)−8%
• Savory snacks (about)−10%

Change in spending within six months of a household starting a GLP-1 medicine. Bar length is the size of the drop, on a 0 to 10% scale. Not plotted because the coverage gives no single figure: sweets and baked goods (a similar large decline), bread, meat and eggs (all lower), and the risers, led by yogurt, with smaller gains reported for fresh fruit, nutrition bars and meat snacks. Source: Hristakeva, Liaukonytė and Feler, Journal of Marketing Research, December 2025, with the Cornell Chronicle summary.

Within six months of starting, grocery spending fell by 5.3% on average. Higher-income households cut by more than 8%. Spending at limited-service restaurants, meaning fast food and coffee shops, fell about 8%. Savory snacks dropped about 10%, and sweets and baked goods fell by similar amounts. Bread, meat and eggs declined too. Yogurt rose the most, followed by fresh fruit, nutrition bars and meat snacks.

Two findings matter for anyone forecasting demand. Among households that kept taking the medicine, the lower spending lasted at least a year, although the gap narrowed over time. And about a third of users stopped during the study window. Their food spending went back to earlier levels, and their baskets ended up slightly less healthy than before they started, partly from more candy and chocolate. See weight regain after stopping for the health side of that pattern.

The authors say they cannot fully separate the medicine's effect from other lifestyle changes people made at the same time. They read appetite suppression as the likely driver, based on clinical trial evidence and on the way spending reversed after people stopped. An earlier working paper reported slightly larger numbers (about 6% overall and nearly 9% for higher earners). That is normal between versions. This page uses the journal figures.

The trial evidence on appetite points the same direction. In a 12-week crossover trial of 30 adults with obesity, weekly semaglutide cut free-choice energy intake at lab meals by 24% against placebo (Blundell and colleagues, 2017). In a smaller 12-week crossover trial of oral semaglutide in 13 evaluable adults with type 2 diabetes, intake fell 38.9% (Gibbons and colleagues, 2020). Those are calories eaten by individuals at a test buffet. The Cornell figure is dollars spent by households at checkout. The two are not interchangeable and should not be converted into one another.

Other data sets agree on direction and disagree on size. In Great Britain, Worldpanel by Numerator found that households with a user spent £418 less than non-user households and bought 299 million fewer packs, which it put at £780 million of lost grocery sales (the period was not specified in the coverage). Chocolate confectionery spending among users fell 18 percentage points more than among non-users. Mouthwash and chewing gum rose 20 and 24 points, which the analysts linked to dry mouth.

Circana's household panel reports that GLP-1 households still outspend non-users on retail food and drink even though they buy less, and it describes a shift toward products offering protein, fiber, energy or hydration. The company projects GLP-1 households will account for 35% of US food and beverage units by 2030. That is a forecast, and the coverage mixes the words "sales" and "units".

Why national snack sales have not collapsed

USDA's Office of the Chief Economist looked at the national picture in February 2026. Snack volume stayed near 10 billion units a year from 2020 through 2024, while GLP-1 prescriptions rose from about 0.5% to just over 2% of adults. The chief economist, Justin Benavidez, said he was surprised the unit data had not moved and that he would revisit it in one to two years. The prescription data stops at 2024 and misses use without a prescription, so it understates use.

Simple arithmetic explains most of the gap. This is an Atlas illustration, not a forecast.

Atlas arithmetic using Circana's household share and Cornell's average effect. These are ceilings, because the Cornell effect narrows over time and not every household member is on the medicine.

Company scenarios land in the same range. Mondelez's chief executive said in February 2026 that the company sees no short-term effect and that a ten-year scenario with 10% to 20% US adoption could affect overall volume by 0.5% to 1.5%. EY-Parthenon has estimated that diet changes tied to GLP-1 use could cost snack makers up to $12 billion in sales over a decade (cited by Reuters). Reuters counted nearly three dozen non-healthcare companies mentioning GLP-1 medicines or weight loss on earnings calls by mid-February 2026, up from 14 a year earlier and five two years before that.

Household size matters too. Reuters, citing a PwC analysis of Numerator data, reported family baskets 4% to 6% smaller and single-person households down as much as 9%. A user's own appetite change is the whole basket in a one-person home and only part of it in a family.

What restaurants are doing, and why the data disagree

The restaurant evidence is split, and the reasons are worth seeing side by side.

Four things can make honest studies disagree here. Receipts record what was bought, while surveys record what people remember. Comparing a person with themselves before and after is different from comparing users with non-users, who may differ in income and habits to begin with. A "restaurant purchase" can be a coffee or a smoothie, which inflates counts for people who snack out. And the sponsors differ: a trade association and an equity research desk both have a stake in how restaurants read the trend. The honest summary is that visits probably fall for many users, spend per visit often rises, and the net effect depends on the restaurant type and the customer's income. Fast-food and quick-service chains show the largest declines in the sources above.

What chains have actually changed

Olive Garden is the clearest real change. Darden tested seven existing dishes in smaller sizes at lower prices, offered alongside the full-size versions. The section reached 40% of restaurants in fiscal Q2 and rolled out fully in January 2026. It costs the company money. Darden said the lighter portions weighed on the check by about 60 basis points in fiscal Q3 (chief financial officer Raj Vennam, March 19, 2026) and added about 50 basis points of mix headwind to the check in fiscal Q1 of 2027 (call of September 24, 2026). On the fiscal Q3 call Vennam also put about 40 basis points of margin investment on the lighter portions menu and delivery fees together. The chief executive, Rick Cardenas, was asked whether demand was value-driven or health-driven and said he does not think it is just GLP-1s, describing a wider wish for smaller portions. In the calls reviewed, Darden put numbers on the cost but not on a sales lift from the menu. It said guests who order the lighter dishes visit more often, and that the choice is strongest at weekend lunch.

The portions are smaller in a specific way. NBC News reported that the chicken parmigiana drops from 1,020 calories and 64 grams of protein to 630 calories and 36 grams. That is 38% fewer calories but 44% less protein, so protein per 1,000 calories slips from about 63 grams to about 57. A smaller plate is not automatically a higher-protein plate.

• Calories, full portion1,020
• Calories, lighter portion630
• Protein, full portion64 g
• Protein, lighter portion36 g

Olive Garden chicken parmigiana, full versus lighter portion. Bar length is relative to the full portion in each pair. The 38% and 44% reductions and the protein per 1,000 calories (about 63 g versus about 57 g) are Atlas calculations from the NBC News figures.

Chipotle launched a High Protein Menu on December 23, 2025, with launch menu items ranging from 15 to 81 grams of protein per item and a snack-ready High Protein Cup of Adobo Chicken or steak. Chipotle listed a national weighted average price of $3.82 for the Adobo Chicken cup at launch, and Axios repeated it. Two entrees are labeled "GLP-1 friendly," and the launch was tied explicitly to the rise of the medicines. Subway added Protein Pockets in January 2026 with more than 20 grams of protein, in 96% of US stores. Its North America president told NBC that GLP-1 use was a key driver. Shake Shack introduced a Good Fit menu.

McDonald's so far describes plans, not a menu. At an investor day on September 23, 2026, the company said it would test grilled chicken sandwiches and wraps, egg bites and bowls aimed at more protein and varied portion sizes (Associated Press). Skye Anderson, president of McDonald's USA, said about 30 million Americans use GLP-1 drugs and are looking for smaller, more protein-packed meals, and that company research shows about 60 million Americans are actively seeking more protein. Both counts are the company's own. The report gives no launch dates and describes no smaller-portion program.

Smoothie King's GLP-1 Support Menu describes high protein, high fiber and zero added sugar. The Associated Press found that the 20-ounce Gladiator GLP-1 Vanilla smoothie carries more calories, sodium and cholesterol than a Krispy Kreme original glazed donut. Both statements can be true at once: a claim about protein and sugar says nothing about total calories.

The National Restaurant Association survey found 49% of users order smaller portions, 43% order more high-protein items, 38% order fewer sweets and 28% of users aged 21 and over order fewer alcoholic drinks. William Blair found nearly half of respondents would dine out more often if smaller-portion options were available, which is the commercial logic behind lighter-portion menus. See the Atlas pages on alcohol on GLP-1 medicines and eating enough.

Supermarkets: the pharmacy is the new data point

Grocers have one advantage restaurants lack: many of them run the pharmacy where the prescription is filled.

Kroger Health launched GLP-1 Complete Support in July 2026, linking pharmacists, registered dietitians, its in-store clinics and a nutrition rating called OptUP, and published shopping guidance built around protein first, hydration, rotisserie chicken, Greek yogurt and portion-controlled snacks. Albertsons Companies posted a GLP-1 series the same month, with a dedicated GLP-1-friendly diet page and a store-tour video, and said a second part would cover how the medicines affect grocery purchases. ShopRite's parent Wakefern offered a limited-time starter kit in February 2026 with samples, coupons and a dietitian's guide for people filling a first prescription. Walmart expanded its Better Care Services platform in April 2026 to connect third-party weight-management providers, some offering nutrition therapy, to its pharmacies. The Healthcare Dive report describes no grocery component.

None of the announcements reviewed says whether prescription data informs food marketing. That is a fair question for readers to keep in mind, and a reason these programs are better read as pharmacy services than as new food products.

British grocers have taken a different route: ranges rather than clinics. A Food Manufacture roundup in January 2026 listed new small-portion or high-protein lines at Morrisons, Marks and Spencer, the Co-op, Asda and Iceland. It noted that GLP-1 wording on the pack is mostly absent, with the ranges described in terms of protein, nutrient density and portion size instead.

What changed besides the packaging: a six-product scorecard

No agreed test exists. Atlas asks three plain questions, using only what the company said and what the nutrition panel shows: did the product change, does the panel back the claim, and does the package use the phrase "GLP-1 friendly"?

Two things stand out. The two products with the most specific nutrition design, Oikos Fusion and Ratio Pro Fiber, keep GLP-1 off the package while still marketing to the group. And Nestlé's own number says most buyers of its GLP-1 companion line are not on the medicines. Atlas reads that as a sign that "GLP-1" works as a marketing wedge into a broader better-for-you market. It is an interpretation, not a finding.

Large manufacturers mostly avoid the phrase. Acosta's Kathy Risch told FoodNavigator that few brands have put it overtly on a package and that she expects more to tread lightly, and consumer research cited in that report suggests the badge can polarize shoppers. Instead, companies talk about protein, fiber and portions. General Mills launched higher-protein Cheerios in December 2024. Coca-Cola raised production of protein-enriched Fairlife milk. PepsiCo plans to test mini-meal options with Sabra and Siete in the US. Conagra's chief executive said on its fiscal third-quarter call (April 2026) that meat snacks grew about 9% in dollars and 10% in volume, and Mondelez's chief executive said its protein bars were growing double digits. Those are company statements for single quarters.

Dietitians quoted in the coverage are wary of the label itself. Suzy Badaracco, a dietitian and food trends consultant, told AP she advises clients against it because an unregulated phrase can lead people to think the food delivers the medicine's benefits. Danone's position, reported by Food Dive, is that calories and sugar matter as much as added protein or fiber. A FleishmanHillard food and beverage executive quoted by the same outlet said the jury is still out on whether the labels work. No study the Atlas found measures whether a "GLP-1 friendly" label changes what people buy.

What the law says about "GLP-1 friendly"

Nothing. There is no federal definition of the phrase.

USDA's Food Safety and Inspection Service approves labels on meat and poultry products. It approved the "GLP-1 Friendly" labels for Conagra and Nestlé because they carry protein and fiber statements and are not misleading, and it told AP there is no regulatory standard for the term. Bloomberg Law reported on July 13, 2026 that, as of June 2, five companies had filed 21 label approval requests for meat or poultry products using GLP-1 language.

FDA oversees most other foods and supplements after they reach the market, without reviewing claims in advance. Kyle Diamantas, who then led FDA's foods program as deputy commissioner, said on March 24 that the agency was aware of the issues and had no intended action. On June 22 an FDA spokesperson said the agency would monitor GLP-1 marketing and act against illegal drug products, an effort that has centered on compounded medicines. The FTC, which polices advertising and requires support for weight-loss claims, declined to comment.

FDA does define some ordinary claims. A "good source" of a nutrient starts at 10% of the Daily Value per serving and "high" starts at 20% (21 CFR 101.54), though meals and main dishes follow separate rules. "GLP-1 friendly" has no such threshold. Courts have not settled the supplement side either: Bloomberg Law reported suits over GLP-1-related supplement marketing, including a case against Grüns that a federal judge dismissed in May while allowing renewed claims, and no court has yet ruled against a company on these claims. See supplements for the Atlas guide.

A one-minute test for any "GLP-1 friendly" label

Use the nutrition panel, not the badge. The ranges below come from dietitians quoted by AP and from USDA guidance, not from a rule that applies to the label. They are starting points to discuss with a prescriber or dietitian, not personal advice.

1. Check the serving size and calories first. Claims usually describe one serving.
2. Look at protein. Dietitians quoted by AP suggest about 20 to 30 grams per meal, or about 1.2 grams per kilogram of body weight per day, to help protect muscle. The protein and strength calculator shows the arithmetic.
3. Look at fiber. USDA's guideline is about 14 grams per 1,000 calories, and constipation is common on the medicines.
4. Check fat, sodium and added sugar. One dietitian in the AP story suggests under 10 grams of fat per serving, because fat can worsen nausea and reflux.
5. Compare with the same brand's standard version. If the panel is identical, the label is the only thing that changed.

A joint advisory from four US nutrition and obesity societies notes that cutting calories can cause nutritional deficiencies and that muscle and bone loss are a risk. That concern is the real basis of the nutrient-dense pitch in most of this marketing. Atlas covers the practical side in what to eat on a GLP-1 medicine, protein and protecting muscle.

The other side of the trade: meal-replacement programs

Some businesses were built around the older way of losing weight, and their results show the same shift from the other direction. Medifast, which sells meal-replacement products through a network of independent coaches, reported 2025 revenue of $385.8 million, down 36.0% from $602.5 million, and a net loss of $18.7 million that includes a $12.1 million non-cash tax charge. Its active earning coaches in the fourth quarter fell 40.6%, from 27,100 to 16,100. The company partly attributed that decline to rapid adoption of GLP-1 medications for weight loss, and guided 2026 revenue to between $270 million and $300 million (company release, February 17, 2026).

This is one company describing its own results, and it names the medicines as one factor among others. It is a data point on how a coach-based meal-replacement model has fared, not a measure of the medicines' effect on food spending.

What changes next

Three things could move the numbers. Adoption keeps rising, and oral options such as the Wegovy pill and Foundayo lower the barrier of an injection. Prices are falling and coverage is widening, which Atlas tracks in the price ledger and the Medicare page. And persistence matters: Cornell found about a third of its users stopped, and Circana reports that half of former users say they are likely to restart, a stated intention rather than observed behavior. A food market that sells to people who start, stop and restart faces a different shopper each year. See weight regain after stopping and eating after stopping.

What we do not know

• Whether the 5.3% grocery effect keeps shrinking beyond a year, or settles.
• Whether restaurant visits rise or fall for typical users. The sources conflict, and none is a full census of receipts.
• Whether protein-labeled foods replace other purchases or add to the basket.
• Whether a "GLP-1 friendly" label changes what anyone buys. No study found.
• How individual calories relate to household dollars. No one has published a direct bridge.

Bottom line

GLP-1 households buy less food, mostly snacks and sweets, and that is measured in real purchase data. National totals have barely moved because the share of households affected is still small, and simple arithmetic puts the ceiling near 1% of grocery dollars at today's adoption. Restaurants are changing portions and prices, and Olive Garden is paying real margin to do it. The "GLP-1 friendly" label is a different matter. It has no legal definition, it sits on products the company says did not change, and the most carefully designed protein and fiber products tend to leave it off. Judge the panel, not the badge.

Questions about food and GLP-1 medicines

How Atlas checked this page

Atlas compared company statements with published nutrition panels where the coverage reported them, and read earnings-call transcripts rather than summaries where available. We did not buy or test products. Figures reported through trade press are labeled "via" and should be checked against the original data set before reuse. This page is educational and is not medical or nutrition advice. Talk with your prescriber or a registered dietitian about your own diet.

Last checked November 7, 2026.