How closely does a food, a meal, or a whole day of eating resemble the kinds of foods linked to longer, healthier lives in research studies? Food Compass scores every food on a single scale from 1 to 100, where higher is better. It looks at nine areas of nutrition all at once, so a sugary breakfast cereal with added vitamins lands in a very different place from a bowl of plain oats.
Score a single packaged product, a whole day of eating, or rank several foods side by side.
Take a photo of the Nutrition Facts panel and ingredient list. The app reads it for you and scores the product.
Pick foods and serving sizes from our catalogue. Score one food on its own or build a list and score the lot together.
See how any food in the catalogue scores in each of the nine areas, plus its processing penalty and NOVA group.
Rank foods side by side. This works across food types, so you can hold a snack bar up against an apple.
The easiest way to score a whole day is to walk through it occasion by occasion in the food diary. Each meal contributes to the day's overall score in proportion to how many calories it adds, so a big bowl of oats counts for more than a small sprinkle of cocoa.
Calorie-weighted. Foods with more calories pull the day's score around more than tiny garnishes, so a single bite of cake will not ruin a healthy day.
Processing-aware. Heavily processed foods take a built-in penalty, with the size of the penalty rising as the level of processing rises.
Three ways to read it. Researcher and policy views open up the detail behind each domain and the methodology behind the score.
The score is not built from one number on the label. It pulls together nine different areas of nutrition, weighing the helpful parts against the harmful ones, so the final number reflects the whole food rather than any single ingredient.
Macro balance, plus key omega and mineral ratios.
How well the food covers the vitamins that matter most for its food group.
How well the food covers the most important minerals for its food group.
The quality of the main ingredients, judged from the first five on the label.
A penalty for added colours, preservatives, and other industrial ingredients.
A penalty based on how processed the food is. Ultra-processed foods take the biggest hit.
The fatty acid profile, looking at saturated, unsaturated, trans, and omega 3 and 6.
How much fibre and protein the food carries per calorie.
Plant compounds with documented health benefits.
Foods are sorted into four NOVA groups based on how industrially processed they are. Processing on its own does not set the score. It nudges the score down by a fixed penalty, and the more processed the food, the larger the penalty.
Fresh fruit, raw meat, milk, dried grains.
Oils, butter, sugar, salt.
Cheese, canned vegetables, simple breads.
Industrial formulations with additives.
A plain-language band, with no methodology jargon. Read it as a comparison between products, not as a personal health verdict.
The full breakdown for each of the nine areas, the processing penalty, and pointers to the methodology behind the score.
Population-level framing for procurement standards, taxation analysis, and food-environment work, with the limits of the underlying study stated plainly.
The catalogue holds 6,719 foods today, drawn from the Canadian Nutrient File and the West African Food Composition Table. Each source keeps its own notes, so any known differences in how foods were measured stay visible. New food composition databases can be added without changing how the scoring works.
Food Compass was first published in Nature Food in 2021 by a team led by Dariush Mozaffarian. It pulled together evidence on which foods tend to support long-term health and which tend not to, and turned that into a single score. A follow-up paper in 2022 in Nature Communications showed that adults in the US who ate higher-scoring diets had lower rates of all-cause mortality, which is the strongest evidence that the score reflects something meaningful about long-term health.
Researcher view links out to both papers if you want to read the original work.