How is GLP-1 adoption affecting packaged food demand?

How many US households have a GLP-1 user?

Between 21% and 23% of US households now include a GLP-1 user, on the trackers that define what they count. PwC's analysis of Numerator data puts households with a current user at 21% in May 2026, up from 9% in January 2025. Numerator's own August 2026 read is 22%. Circana counts 23% of households with at least one user over a year, as of fall 2025.

The adult rate is roughly half that. KFF's November 2025 poll found 12% of adults currently taking a GLP-1 for weight loss or a chronic condition; Gallup's mid-2026 panel found 11% taking one for weight loss alone, up from 3% in 2024. One user puts a whole household in the count, and the household buys the groceries. This means the household rate is the base for any grocery model.

The outlier is NIQ's 4.5%: GLP-1 users represent 4.5% of households yet drove 43% of sales growth for products with muscle health claims, across two 52-week periods in 2024 and 2025. NIQ does not say how a user household is identified, and its trip data covers weight-loss users only. The point it carries is real, a small group driving a large share of protein-led growth, but without a definition it is not market data you can cite as a penetration rate.

Read that way, the 4.5% and the 23% answer different questions. Before you reconcile estimates that don't match, establish whether they count the same thing.

How much less do GLP-1 households spend on food?

5.3% within six months is the best-measured effect in the evidence. A Cornell team led by Sylvia Hristakeva and Jura Liaukonyte linked repeated surveys on GLP-1 use to purchase records from Numerator's panel of about 150,000 US households (Journal of Marketing Research, December 2025). Grocery spending falls an average of 5.3% within six months of someone in the household starting the drug, and more than 8% in higher-income households. Spending at fast-food chains, coffee shops and other limited-service restaurants falls about 8%.

Numerator's August 2026 update puts the grocery decline at 3.8%, but gives neither the window nor the comparison group, so it neither confirms nor contradicts the 5.3%.

The awkward fact is that users still outspend non-users. NIQ finds GLP-1 users spent 15.1% more per trip and 7.7% more over the year than nonusers in the 52 weeks to August 2, 2025, even as weight-loss users made 6.4% fewer trips; Circana reports the same pattern. Both hold at once. The 5.3% compares a household with its own past; the 7.7% compares two groups that differed before anyone took a drug: 55% of US GLP-1 users live in households earning more than $100,000 a year.

This means a category model applies the change within adopting households to their baseline spending. The gap between users and non-users describes who the users are, not what the drug does.

Which packaged food categories lose, and which gain?

The cut is concentrated in snacks and sweets. In the Cornell analysis, savory snack spending falls about 10% after adoption, with similarly large decreases in sweets, baked goods and cookies; bread, meat and eggs also decline. Yogurt gains most, followed by fresh fruit, nutrition bars and meat snacks. Because part of the money moves to fresh fruit and yogurt, packaged snack and bakery categories carry more than the 5.3% average.

Numerator saw the same shape in November 2023: GLP-1 households cut food spending 1.6% in the first nine months of that year against 0.5% for other households, with packaged sweet bakery, chips and cookies underperforming and soft drinks losing ground to sports and energy drinks.

This leaves a portfolio question, not a market-size question. A savory snack business loses roughly twice the average basket cut in every adopting household; a yogurt or protein-bar business draws a tailwind from the same households.

Is Circana's 35% a share of sales or of units?

Units, on Circana's own page. The November 18, 2025 release is headlined as 35% of US food and beverage sales by 2030; its text says GLP-1 households are projected to represent 35% of all food and beverage units sold. Both versions circulate; the second is the one Circana defines.

The difference matters to a category manager. A unit share equals a dollar share only if the group pays the same average price as everyone else, and NIQ's 15.1% higher spend per trip says it does not. Model the 35% as a share of units, and convert it to dollars only with the price-mix assumption stated beside it.

Nor is it a forecast of lost demand. It means more than a third of units will be bought by households with smaller appetites and a stronger pull toward protein; the volume effect on a category still comes from the change per household. This means the 35% belongs in the assortment and pack-size discussion, not in the top-line forecast.

How big is the hit to industry volume and revenue?

About 1 to 2 percentage points of US food and beverage volume, on a top-down estimate. OC&C Strategy Consultants derived it in February 2025 from roughly 9% of adults on a GLP-1 and an average calorie cut of 15% to 30% per user. Adult use has since risen to 11% to 12%, so the same arithmetic now moves up in proportion.

A bottom-up route lands in the same band. A 5.3% cut in the grocery bills of roughly one household in five is about 1% of US grocery spending: a little more once users' higher spending is counted, a little less once the third who stop are taken out. It is the arithmetic you would use to size a market top-down and then check from the bottom, one route from calories and one from transactions.

The forward numbers are larger and softer. J.P. Morgan projects an annual food and beverage revenue reduction of $30 billion to $55 billion by 2030 to 2034, with consumers taking in 21% fewer calories and spending 31% less on groceries: assumptions inside a forecast, not measurements. The same page puts the 2030 US user base at about 25 million in its key takeaways and 30 million in its text, a gap of a fifth on the input that drives the band.

None of these converts into another one for one. For the industry, the near-term planning number is a low single-digit volume headwind; for savory snacks, roughly a tenth of each adopting household's spending in the category. The 2030 revenue band is a stress test, not a base case.

What should a category plan assume?

For a client deck, the GLP-1 impact on food industry demand reduces to four inputs, each with its own source: household penetration of 21% to 23% on a stated basis; the change within adopting households, 5.3% of grocery spending in six months and about 10% in savory snacks; the share of users who stop, about a third in the Cornell study; and the unit share those households hold by 2030, 35% on Circana's projection.

Persistence is the open variable. Households that stopped the drug in the Cornell study returned to pre-adoption food spending; Circana finds produce growth survives discontinuation while beverages and refrigerated and frozen food rebound. Caspr's Market Research Study on the subject reads this as a structural trend with cyclical amplitude: the direction is durable, the speed moves with coverage, price and discontinuation.

Manufacturers are planning on that basis. PepsiCo launched Propel Clear Protein in May 2026, with 20g of protein and 90 calories a packet, developed with GLP-1 users in mind; Conagra badges Healthy Choice meals it markets as GLP-1 friendly. The client question is no longer whether to respond but which part of the portfolio to defend.

The published figures

Why the figures differ

• definitional NIQ's 4.5% and Circana's 23% of US households NIQ's 4.5% is the user-household share in a sales-growth comparison, with no published user definition; Circana counts every household with any user over a year.
• definitional PwC's 21% and Circana's 23% of US households A point-in-time count of households with a current user against a count of households with any user over a full year, which reads higher.
• definitional KFF's 12% and Gallup's 11% of US adults KFF counts current use for weight loss or a chronic condition in late 2025; Gallup counts use for weight loss only in mid-2026.
• definitional Circana's headline (35% of sales) and its text (35% of units sold) One projection in two units. Unit and dollar shares diverge when a group's price mix differs, and users spend 15.1% more per trip.
• definitional Cornell's 5.3% fall in grocery spending and NIQ's 7.7% higher annual spend by users A change within the same households over time against a level gap between two groups that differ in income before treatment.
• genuine J.P. Morgan's 25 million and 30 million US GLP-1 users in 2030 Same page, same quantity, same year: about 25 million in the key takeaways, 30 million in the text, unreconciled.

Frequently asked

Keep reading

• Why battery storage cost forecasts disagree
• SaaS revenue multiples: NTM against TTM