Why each figure has a lens on it
Every chart below reports an aggregate, and every aggregate here is misleading in the same specific way: the category behaves nothing like the things inside it.
So each figure carries a magic lens, a circle you drag across the chart. Under the glass, the same data re‑renders at a finer grain; outside it, nothing moves. You can look closely at one product without losing the comparison you were making. If you want the interaction itself explained, I built a piece about it. Here it is just the instrument.
Drag any lens · drag the small circle at its lower‑right to resize · or focus it and use arrow keys, + / −, 0. Every figure also ships its numbers as a plain table for screen readers and Cmd‑F.
A real medical reason, and a much wider market
The demand is real. What gets oversold is how many people it applies to.
Start with the part that is well evidenced. GLP‑1 drugs produce fast, large weight loss, and a meaningful share of what comes off is muscle. A 2025 meta‑analysis of 22 randomised trials (n = 2,258) found GLP‑1 receptor agonists reduced lean mass by 0.86 kg, about a quarter of total weight lost. An expert review the same year put it more plainly: trial participants lost “10% or more of their muscle mass” over 68–72 weeks, which the authors called “approximately equivalent to 20 years of age‑related muscle loss.” As of May 2024, 12% of US adults told KFF they had ever taken one.
For those people — plus older adults, who geriatric nutrition consensus has said since 2013 need 1.0–1.2 g/kg/day rather than the 0.8 RDA — more protein is sound advice.
Now the part that gets less attention. The best published analysis of what Americans actually eat — NHANES 2001–2014, n = 57,980 — found that every demographic group’s mean protein intake exceeds the Estimated Average Requirement, that adults 19–30 average 88 g a day, and that fewer than 1% of any group approaches the upper end of the acceptable range. The authors’ conclusion: “protein intake, as a percentage of energy intake, is not excessive in the American diet.”
Even adults over 71 average 1.10 g/kg — inside the expert range, if only just. So the summary is not that everyone already gets enough. It is that the people who really need more protein can be picked out, and they are not most shoppers.
Shoppers, by their own account, are mostly guessing. In IFIC’s 2025 survey 71% said they were trying to consume protein — and 79% said they did not know or were not sure how much they need.
Sources. Karakasis et al., Metabolism 2025;164:156113 (PMID 39719170) · Mechanick et al., Obesity Reviews 2025;26(1):e13841 (PMID 39295512) · Wilding et al., NEJM 2021;384:989 (STEP 1) · Bauer et al. (PROT‑AGE), JAMDA 2013;14(8):542 · Berryman et al., Am J Clin Nutr 2018;108(2):405 (PMID 29931213) · KFF Health Tracking Poll, May 2024 · IFIC 2025 Food & Health Survey (industry‑funded trade association — weigh accordingly).
Two million labels, and what is left after filtering
USDA publishes its entire Branded Foods database. I took the April 30, 2026 release — 1,999,950 products — and kept the ones that put the word protein in their own name. That is the marketing claim, stated by the manufacturer, and it is reproducible by anyone.
The first thing I found was about the labels rather than the food.
Of 31,293 US protein‑named products, 91.6% declare protein — but only 35.2% declare added sugar at all, and 10.6% declare sugar alcohols. For roughly two‑thirds of the shelf, the tradeoff this report is about cannot be evaluated from the public data. Everything that follows is drawn from the minority of products that told us.
After keeping only products declaring both protein and added sugar, deduplicating repeat submissions, and dropping records with physically impossible values — 1.0% of them, including Premier Protein shakes logged at 272 g of protein per 100 ml — 3,578 products remain across six form factors.
Source & method. USDA FoodData Central, Branded Foods full download, 2026‑04‑30 release. Per‑serving values derived as per_100 × serving_size/100; rows failing physical bounds (any macronutrient >100 g/100 g, energy >900 kcal/100 g) excluded. Manufacturer‑submitted label transcriptions, not independent assays. Counts are of products, not sales — this is what is on the shelf, not what is in the basket.
The sugar is not evenly distributed
Mean added sugar per serving runs from 1.3 g in protein powder to 8.6 g in protein cereal — a 6.6× spread across products that make the same basic promise.
But an average hides whatever is unusual inside it. Drag the lens across a bar and it stops reporting the average: it shows you the full distribution inside that category, decile by decile, from its cleanest tenth to its worst.
The bars are category means; under the glass they become the distribution. Watch Bars in particular — a mean of 4.8 g looks moderate, and the top tenth of the category starts at 12 g. Powder is the one tight category — its median is zero.
Notice that it runs backwards. The two categories with the most protein — powder at 21.0 g and shakes at 18.6 g per serving — sit at opposite ends of the sugar range, while cereal delivers the least protein (11.1 g) alongside the most sugar. Measured as grams of added sugar per 10 g of protein, cereal is 7.7 and powder is 0.6 — a thirteenfold gap for the same nutrient.
FDA publishes its own threshold — “20% DV or more is a HIGH source of added sugars”, which on a 50 g Daily Value means 10 g a serving. By FDA’s own definition, 36.4% of protein cereals are HIGH in added sugar, against 1.2% of powders.
Where the claims and the sugar land
Every dot is one product: protein per serving across, added sugar per serving up. The two lines are FDA’s own thresholds — 20 g of protein to call yourself an “excellent source” under 21 CFR 101.54, and 10 g of added sugar to be “high.”
A scatter this dense cannot be labelled — 1,459 points will not carry 1,459 names. That is what the lens is for. Drag it anywhere and the cloud underneath resolves into named products you can look up.
Take the lens to the upper left — low protein, high sugar — and read what is there. These are products sold on a protein claim that deliver less protein than a cup of yogurt alongside more added sugar than FDA calls high. The lower right is the quadrant the claim implies: real protein, little sugar, and it is almost entirely powders and shakes.
Aggregated, the shelf sorts like this: of 3,578 products carrying a protein name, only 1,301 (36.4%) actually clear FDA’s “excellent source of protein” bar, while 535 (15.0%) are high in added sugar. The single largest group — 1,864 products — is neither: low protein and low sugar. Most of this aisle is an ordinary snack with protein written on the front.
Each category’s share of the four quadrants. Powder is the clear exception — 75.8% of powders are high‑protein/low‑sugar, against 8.5% of protein cereals.
Thresholds. 21 CFR 101.54(b)(1): “high,” “rich in” or “excellent source of” requires 20 percent or more of the DRV per reference amount; 21 CFR 101.9(c)(7)(iii) sets the protein DRV at 50 g, so the bar is 20 g. Added‑sugar DV of 50 g and the 5%/20% low/high rule of thumb are FDA’s, from its Added Sugars labelling guidance. Verified against live eCFR text.
Two products, the same claim, a 28‑gram difference
None of this breaks a labelling rule, which is what makes it worth looking at.
A bar with 20 g of protein and no added sugar and a bar with 20 g of protein and 28 g of added sugar make the identical front‑of‑pack protein claim, because the claim is only about the protein. The sugar is disclosed truthfully on the back, in a panel most people do not compare across products in a store.
There is a second gap, less obvious. Under 21 CFR 101.9(c)(7)(ii), the protein %DV must be calculated on the amount “multiplied by the amino acid score corrected for protein digestibility” — the PDCAAS correction — while the gram number on the front is uncorrected. A plant protein declaring 20 g with a digestibility score of 0.70 is 14 g of usable protein: 28% DV, not 40%. And the %DV is voluntary unless a formal nutrient content claim is made. A brand can print a big gram number and omit the one figure that would reveal the correction.
The ten highest‑sugar products sold under a protein name
Filtered to products with at least 8 g of protein, so these are not trace amounts. Drag the lens down the rows to expand each into its full panel.
Every row carries its FoodData Central ID, so each is independently checkable at fdc.nal.usda.gov. Bolthouse Farms’ “Perfectly Protein” line occupies half the list, and a strawberry protein toaster pastry tops it at 48 g of added sugar against 12 g of protein — 96% of the entire daily added‑sugar value in one pastry. Quaker’s cookies‑and‑creme bar makes the list with a legitimate 20 g of protein and 29 g of sugar alongside it.
Protein food is growing. Diet programmes are not.
Market‑research growth figures for “the protein market” disagree with each other by factors of several, because no two firms define the category the same way. So I used audited filings instead. They cover less ground, but they have been checked.
Three companies, three fiscal calendars, one decade. BellRing (Premier Protein, Dymatize) is effectively a protein pure‑play. Simply Good Foods owns Quest, Atkins and OWYN. Medifast sells much the same thing, packaged as a weight‑loss programme.
Indexed to each company’s own FY2019 = 100. Take the lens to the Simply Good line and it decomposes into its brands — The company total is flat; the brands underneath it are not.
BellRing compounded net sales at 15.8% a year for seven years, to $2.32 bn in FY2025. Simply Good grew +9.0% in FY2025 — underneath which Quest compounded +17.5%/yr while Atkins declined −4.5%/yr, falling 14.5% in FY2025 alone. Quest has gone from 47% of North American sales to 61%. Management is explicit about the substitution: they are “working with retailers to replace lower performing Atkins products with higher performing Quest and OWYN products.”
And Medifast lost 75.9% of its revenue from its FY2022 peak — $1.60 bn to $386 m — with GLP‑1s named in the MD&A as a cause.
The growth is volume, not price
This is the most useful of the three disclosures, and BellRing publishes it every year. In FY2022 — the inflation year — Premier Protein sales rose 7% while volumes fell 8%. That is price growth rather than demand growth. From FY2024 the sign flips: volume growth exceeds dollar growth, meaning real per‑unit prices fell while units surged. FY2025: sales +17% on +15% volume for Premier, +13% on +23% volume for Dymatize.
Headline sales growth per year; under the lens, the volume figure quoted in the same sentence of the same filing. Where the volume bar exceeds the sales bar, unit prices fell. So the recent growth looks like real demand rather than higher prices.
GLP‑1s, as described by the companies they affect
All three passages below are from audited 10‑K filings covering overlapping twelve‑month periods. They describe the same drug class.
“We expect the convergence of several factors to support the continued growth of the convenient nutrition category, including: consumers’ increasingly dedicated pursuit of active lifestyles and growing interest in nutrition and wellness (including the use of GLP‑1 medication)…”
“Our brands can appeal to… consumers looking to manage weight through the several approaches including using medications such as GLP‑1s… consumers are increasingly looking to increase the level of protein consumption in their diets.”
“Revenue decreased $216.7 million, or 36.0%… The decrease in the number of active earning coaches was driven by continued pressure with client acquisition reflecting broader challenges in the operating environment, including rapid adoption of GLP‑1 medications for weight loss.”
Medifast responded by moving toward protein itself, launching mini meals with “20 grams or more of high‑quality protein to help retain lean muscle mass” and an explicit “GLP‑1 Nutrition Support Plan.” The company the drug hurt most is now selling protein to the people taking it.
Sources. All figures in §07–08 from SEC EDGAR: XBRL company‑concept API for consolidated net sales; brand‑level dollars from Simply Good’s audited segment note; BellRing brand growth and volume percentages quoted verbatim from MD&A. BellRing CIK 1772016 · Simply Good Foods CIK 1702744 · Medifast CIK 910329. Year‑over‑year and CAGR figures are my arithmetic on directly tagged values. BellRing does not disclose brand‑level dollar levels in its 10‑K, so none are shown.
Three things worth taking away
Not advice about how much protein you need — I am not qualified to say, and it depends on you. These are just a few things to look for on the shelf.
- According to the FDA, twenty grams per serving is the line. Below 20 g per serving a product cannot legally call itself an excellent source of protein — but it can still put “protein” in its name. Nearly two‑thirds of this aisle is on the wrong side of that line.
- Read the added‑sugar row, not the total‑sugar row. They are different lines on the panel and have been since 2020. Ten grams is FDA’s “high” threshold. A third of protein cereals cross it.
- Treat a missing added‑sugar figure as a warning sign. Two‑thirds of protein‑named products in the public database do not declare one. A missing number does not mean there is no sugar.
And the one that matters most: the category average is almost never about the thing you are holding. Protein bars average 4.8 g of added sugar; the one in your hand might have 28. That is why every figure here has a lens on it.
What this report is not
- These are labels, not lab tests. FoodData Central transcribes what manufacturers submit. Nobody put these products in a mass spectrometer, including me.
- Counts are of products, not sales. Every count is a count of SKUs on the shelf, unweighted by how many anyone buys. A category mean here is the mean of this sample, not of the market.
- The universe is a naming convention. I selected products with “protein” in the product name. That is reproducible and it is what the marketing claim looks like — but it under‑counts real high‑protein foods that don’t say so, which is why protein yogurt is thin here (n = 70) and protein ice cream — at n = 15 — was too sparse to chart at all. Halo Top and Yasso do not put the word in their names.
- Protein gummies are absent. I went looking, because they are the form factor most often cited as the emblem of the trend. FoodData Central has almost nothing: a handful of stale 2019–2021 records, none declaring added sugars. I would rather report the gap than extrapolate across it.
- No category‑level market size appears anywhere above. Published “protein market” figures come from paid research whose methodologies are not public and which disagree with each other by multiples. I used audited filings instead, which cover three companies rather than a market.
- The KFF 12% figure is as of May 2024 and is certainly higher now. I have not updated it, because I could not verify a newer number to the same standard.
- Correlation, throughout. Nothing here demonstrates that GLP‑1 adoption caused protein demand. The strongest claim the evidence supports is that three companies exposed to the same drug class, filing under penalty of perjury within twelve months of each other, describe it as two tailwinds and one 36% revenue decline.
Sources
None of these numbers are mine. Each one comes from a source below.
Data
- USDA FoodData Central, Branded Foods full download, 2026‑04‑30 release (1,999,950 products). fdc.nal.usda.gov
- SEC EDGAR — XBRL company facts and 10‑K filings. BellRing Brands (CIK 1772016), Simply Good Foods (CIK 1702744), Medifast (CIK 910329).
Regulation
- 21 CFR 101.54(b)(1) — “high/rich in/excellent source of” = ≥20% DV per RACC.
- 21 CFR 101.9(c)(7) — protein DRV of 50 g; (c)(7)(ii) the PDCAAS correction for %DV; (c)(7)(i) %DV voluntary absent a claim.
- 21 CFR 101.9(c)(6)(iii) — mandatory added‑sugars declaration. FDA Added Sugars labelling guidance — 5% DV low / 20% DV high.
- All verified against live eCFR text, not secondary summaries.
Science
- Berryman CE et al. “Protein intake trends and conformity with the DRIs…” Am J Clin Nutr 2018;108(2):405–413. PMID 29931213. (NHANES 2001–2014, n = 57,980.)
- Karakasis P et al. “Effect of GLP‑1RA on lean body mass…” Metabolism 2025;164:156113. PMID 39719170. (22 RCTs, n = 2,258.)
- Mechanick JI et al. “Nutritional considerations in patients treated with GLP‑1RAs.” Obesity Reviews 2025;26(1):e13841. PMID 39295512. Industry‑funded authorship — see §02.
- Wilding JPH et al. “Once‑weekly semaglutide in adults with overweight or obesity.” NEJM 2021;384:989–1002. (STEP 1.)
- Bauer J et al. (PROT‑AGE Study Group). JAMDA 2013;14(8):542–559. (1.0–1.2 g/kg/day for older adults.)
- Institute of Medicine, Dietary Reference Intakes for Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol, Protein, and Amino Acids, 2005. (RDA 0.8 g/kg; AMDR 10–35% of energy.)
Survey
- KFF Health Tracking Poll, May 2024 — GLP‑1 ever‑use among US adults.
- IFIC Food & Health Survey, 2025 — protein‑seeking behaviour. Published by an industry trade association; treat as directional.
Analysis, charts and lens by Carey Chou. The derivation is a handful of Python scripts over a public database; anyone with the same download gets the same numbers. If you find an error, it is mine — get in touch.