Front-of-pack claims vs nutrition: 145,609 foods tested

A large observational study published online in Nature Food on September 24, 2026, concludes that on-package claims on U.S. packaged foods and beverages are imperfect signals of overall nutritional quality. Researchers from Cornell University analyzed 145,609 newly launched products and found that while over half carried at least one front-of-package claim, approximately two-thirds of all evaluated products were classified as less healthy under a standardized nutrient profiling model. For some claims, the products carrying them scored worse overall than comparable products without them; for example, "low sodium" claims on beverages were associated with higher total sugar.

Key takeaways

• Over half of new U.S. food and beverage launches from 2015 to 2024 carried at least one claim (2.8 on average among products with claims), and about two-thirds of all products were classed as less healthy under the UK Nutrient Profiling Model (NPM).
• In beverages, "Low/No/Reduced Sodium" and "Low/No/Reduced Saturated Fat" claims were associated with higher total sugar content.
• In food products, "Low/No/Reduced Trans Fat" claims were associated with higher saturated fat.
• "No Added/Artificial Flavourings," "No Additives/Preservatives" and "No Added/Artificial Colourings" claims showed a discrepancy within some categories, including condiments and meals and side dishes.
• The authors say the discrepancies "do not imply regulatory violations." A single claim describes one attribute; the Nutrition Facts panel shows the rest.

What happened

On September 24, 2026, researchers Pei Zhou, Yuqing Zheng, Harry M. Kaiser, and Miguel I. Gómez at the Charles H. Dyson School of Applied Economics and Management at Cornell University published an open-access study in Nature Food titled "On-package claims are imperfect signals of overall nutritional quality." Cornell also described the findings in a press release titled "What 'healthy' labeling on packaged food conceals".

The investigators examined product-level packaging and formulation records from the Mintel Global New Products Database spanning a ten-year window from January 2015 through December 2024. The final dataset comprised 145,609 food and beverage products launched in the United States, covering beverages and 10 food categories of processed consumer packaged goods; fresh produce, baby food and other non-comparable groups were excluded.

Funding for the investigation was provided in part by the United States Department of Agriculture (USDA) National Institute of Food and Agriculture (NIFA) grant 2024-67023-42730, alongside the Cornell Food Industry Management Program, the Cornell Atkinson Center for Sustainability, and the Cornell SC Johnson College of Business. The authors declared no competing interests.

What changed

Rather than looking at consumer health outcomes or shopper buying habits, the study is an observational, descriptive analysis of product data: it compares the claims on each product with a nutrient-profile score. The researchers scored each product's nutritional profile using the established UK Nutrient Profiling Model (NPM 2004-2005). Under this system, foods with an NP score of 4 or greater and beverages with a score of 1 or greater are categorized as "less healthy."

The study categorized product claims into four distinct Mintel classifications:

• Minus claims: Reductions in nutrients to limit, such as "Low/No/Reduced Calorie," "No Added Sugar," "Low/No/Reduced Trans Fat," or "Low/No/Reduced Sodium."
• Plus claims: Additions of beneficial components, such as "High/Added Fibre" or "Vitamin/Mineral Fortified."
• Natural claims: Processing and ingredient attributes, including "Organic," "Whole Grain," "GMO Free," "No Added/Artificial Flavourings," and "No Additives/Preservatives."
• Functional claims: Targeted physiological callouts, such as "Cardiovascular Health" or "Antioxidants."

Using two-sided Wilcoxon rank-sum tests to compare the scores of products with and without each claim, the authors report that "for food, most claims are aligned, except for Low/No/Reduced Trans Fat." "For beverages, patterns are more heterogeneous, with four claims exhibiting discrepancy: Low/No/Reduced Calorie, Sugar Free, Low/No/Reduced Sodium and Low/No/Reduced Saturated Fat." On trade-offs, they write that "in beverages, Low/No/Reduced Sodium and Low/No/Reduced Saturated Fat claims are associated with higher total sugar content, while in food products, Low/No/Reduced Trans Fat claims are associated with higher saturated fat."

Who is affected

The findings concern what shoppers see on new products. Claims "remained pervasive throughout the decade": over half of all new products carried at least one claim, reaching a peak of 61.1% during 2019 and 2020 before resting at 54.5% in 2024. Among products that displayed any claims, the average count was 2.8 claims per package.

Despite this, about two-thirds of all products in the dataset were categorized as less healthy under the NPM. The share classed as less healthy varied by category:

• Candy: 93.24% classified as less healthy
• Snacks: 90.38% classified as less healthy
• Desserts and Ice Cream: 83.79% classified as less healthy
• Bakery: 80.76% classified as less healthy
• Beverages: 32.36% classified as less healthy
• Processed Fruit and Vegetables: 2.18% classified as less healthy

Within beverages, the sodium-claim discrepancy arose in carbonated soft drinks (439 products with a sodium claim versus 1,802 without), fruit or flavored still drinks, and ready-to-drink coffee and tea, and was "driven mainly by higher total sugar." For trans-fat claims, higher scores in Dairy and Eggs "reflect lower protein and higher saturated fat, energy and sodium," and in Meals and Side Dishes "higher sodium and saturated fat."

What it means for shoppers

For shoppers, the practical point is that one claim describes one attribute, not the whole product. In this dataset, food products with trans-fat claims were associated with higher saturated fat, even though the authors stress this does not mean the claims broke any rule.

Natural-type claims work the same way. As an example of the claim type (not a product the study analysed), Lay's Oven Baked Original carries a "No artificial flavors" claim according to its Open Food Facts listing, which gives the ingredients as dried potatoes, corn starch, corn oil, sugar, sea salt, soy lecithin, dextrose and annatto extracts. The claim tells you about flavorings; it says nothing about the rest of the nutrition profile.

A few practical checks:

• Turn the package around to check the Nutrition Facts panel for saturated fat, sodium, and added sugars.
• Review the complete ingredient deck rather than relying on front-of-pack buzzwords like "organic" or "natural."
• Check whether a reduction in one nutrient has been balanced by an increase in another, such as elevated sugar in low-sodium beverages.
• Remember that even aligned claims are not a guarantee: the authors write that "many products with aligned claims still exceed the NPM thresholds," naming "Vitamin/Mineral Fortified, High/Added Fibre and Probiotic/Prebiotic" as limited exceptions.

The background

In the United States, nutrient content claims must be authorized by Food and Drug Administration (FDA) regulations. According to FDA, nutrient content claims describe the level of a nutrient in a product using terms such as "free," "high," and "low," or compare nutrient levels using terms like "more," "reduced," and "lite." As the Cornell researchers note, "these discrepancy patterns do not imply regulatory violations," and the UK model they used "is not directly applied to on-package claim regulation in the United States."

The FDA announced a final rule on December 19, 2024, updating the "healthy" claim. It requires a food to contain a certain amount from at least one recommended food group "and meet specific limits for added sugars, saturated fat and sodium," which makes it a profile-based claim rather than a single-nutrient one.

For a broader walkthrough of the back of the pack, see How to Read a Food Label: The Ingredient List, Line by Line. For a European score tested against real products, see Nutri-Score 2023 tested on 408,399 EU food labels, and for a state approach to a front-of-pack seal, see California Creates a 'Non-Ultraprocessed Certified' Food Seal.

What to watch next

Several research limitations must be kept in mind when evaluating the Cornell data. The analysis is strictly observational and descriptive; in the authors' words it "does not directly capture consumer responses to on-package claims," and it did not measure diets or health outcomes. Data on newly launched products "may not fully represent the broader food supply or products most commonly purchased by consumers." The paper does not name brands in the dataset.

The paper does not say how its results would change under FDA's proposed front-of-package "Nutrition Info" box, which is not final; see FDA's Front-of-Package Nutrition Box Is Still Not Final.

To see what a product's ingredients are and what regulators say about each one, scan your next packaged grocery item with the free Aurascan app (iOS beta on TestFlight; waitlist for other platforms), which explains each ingredient with cited evidence.