What Nutri-Score and NOVA actually tell you (and what they don't)

A product can carry a Nutri-Score of A and a NOVA grade of 4 at the same time — a diet soda is the classic example. Neither number is wrong. They’re just answering different questions, and most people reading a grocery label have no idea they’re looking at two separate scoring systems layered on top of each other. Once you know what each one is actually measuring, both become useful instead of confusing.

I added both to Stocky because scanning a barcode and getting one flat “healthy/unhealthy” verdict would have been dishonest — the real picture needs two axes, not one. Here’s how to read them.

Nutri-Score answers “what’s in it”

Nutri-Score is a nutrient-density grade, A through E, computed from a fixed formula: it rewards fiber, protein, and the proportion of fruit, vegetables, and legumes, and penalizes energy density, saturated fat, sugar, and salt. It’s a per-100g calculation, which matters — it’s comparing nutritional composition, not serving size or how the food got made.

That’s exactly why it’s good at what it’s good at: comparing two similar products on the same shelf. Two breakfast cereals, two yogurts, two loaves of bread — Nutri-Score is a fast, honest way to see which one has less added sugar and salt for the same amount of food. It was designed for that specific job, and it does it well.

It’s also why it breaks down outside that job. Nutri-Score doesn’t know or care whether a product is a whole food or an industrial formulation, so a diet soft drink with no sugar and no fat can score better than plain olive oil, which is calorie-dense by nature even though it’s minimally processed. The formula sees energy density and penalizes it; it has no concept of “this is one ingredient” versus “this is fourteen.” That’s not a bug in the score — it’s a boundary. Nutri-Score was never trying to answer the processing question.

NOVA answers “what was done to it”

NOVA classifies food into four groups by how much industrial processing and formulation it went through, independent of its nutrient profile:

• Group 1 — Unprocessed or minimally processed. Fresh, dried, or frozen fruit and vegetables, plain grains, eggs, plain yogurt. Nothing added, nothing extracted, no new substances created.
• Group 2 — Processed culinary ingredients. Oils, butter, sugar, salt — things extracted from Group 1 foods, used to cook with rather than eaten alone.
• Group 3 — Processed foods. Group 1 or 2 foods with something added to them — canned vegetables in brine, cheese, freshly made bread, canned fish in oil. Usually two or three recognizable ingredients.
• Group 4 — Ultra-processed food. Industrial formulations with ingredients you wouldn’t find in a home kitchen — hydrogenated oils, protein isolates, flavor compounds, emulsifiers, artificial colors. Built to be shelf-stable, intensely flavored, and ready to eat with minimal preparation.

NOVA is where the “healthy-looking but heavily engineered” products get caught. A cereal bar can carry a Nutri-Score of B — reasonable sugar, decent fiber, low saturated fat — and still be NOVA 4, built from isolated fibers and flavoring agents rather than oats and dried fruit. The nutrient math checks out; the ingredient list tells a different story. That gap is the whole reason to look at both.

Reading them together

The two scores form a 2×2 that’s more informative than either alone:

• A/B + NOVA 1–2 — the safe default. Whole or lightly processed food with a good nutrient profile. Plain oats, fresh produce, unsweetened dairy.
• A/B + NOVA 4 — the one worth double-checking. Nutrient math looks fine, but it’s a formulation, not a food — check the ingredient list before assuming it’s a staple.
• D/E + NOVA 1–2 — often something dense and traditional rather than something to avoid outright: olive oil, cheese, nuts. High energy density by nature, not a processing problem.
• D/E + NOVA 4 — the actual category to eat less of. Poor nutrient profile and heavily engineered. This is what both systems are, separately, warning you about.

The practical rule: use Nutri-Score to pick between near-identical products on the same shelf, and use NOVA to catch the ones where a good nutrient grade is doing the disguising. Neither score tells you a food is forbidden — a D-graded block of cheese isn’t the enemy, and an A-graded diet drink isn’t automatically a good choice just because the letter looks green. They’re inputs, not verdicts.

Where the scores come from

Both grades exist on products in the Open Food Facts database, computed from the ingredient list and nutrition table the manufacturer or the community submitted — not something an app can quietly bias in either direction. Coverage is the real limitation, not accuracy: a well-known packaged product usually has both grades already computed, while a small local brand or something homemade obviously won’t carry either, because there’s no submitted label to compute them from. When Stocky shows a Nutri-Score or NOVA grade next to a scanned item, that’s what it’s surfacing — nothing modeled, nothing app-specific, just the published grade attached to that barcode.

The takeaway that’s worth keeping past this post: don’t collapse two questions into one label. “Is this nutritionally reasonable for its category” and “how much was this engineered before it reached the shelf” are genuinely different things to know about food, and a single grade — however convenient — can’t answer both at once.