Most food-scoring apps hide the rubric. We don't. Here's exactly what goes into a NutraSafe grade, what we deliberately leave out, and why we drew the lines where we did.
A NutraSafe grade is additive risk only. We classify every additive in a product into a risk tier, dedupe related additives so the same concern isn't counted twice, and drop a grade the more risk is present. Nutrition, processing level and our plain-English verdict are shown separately — not mixed into the grade.
Every UK product we've seen gets one letter — A, B, C, D or E. The letter answers one question: how much additive risk is in this product?
1. Risk tier per additive. Every additive gets classified into high, moderate or low risk based on published research and regulatory status (FSA, EFSA). A colouring linked to behavioural effects in children isn't treated the same as an acidity regulator.
2. Deduplication by function. If a product has three nitrate variants, that's one concern — at the highest risk level — not three. Same for multiple sweeteners, multiple phosphates, multiple sulphites. Counting related additives separately would punish a product for the way it spells its ingredients list.
3. NOVA 4 upgrades. Some ingredients aren't "additives" by name but are reliable markers of ultra-processing — modified starches, protein isolates, hydrogenated fats, glucose syrups. When we see them, we upgrade the processing signal.
4. Functional ingredient exclusion. Added vitamins, minerals and nootropics don't pull the grade down. A cereal fortified with B12 and iron shouldn't score worse than the same cereal without them. Fortification is useful, not a penalty.
These are all worth knowing — they're just not what the letter is measuring. We show them elsewhere in the app.
Earlier versions of our grade tried to bundle nutrition, processing and additives into a single number. It gave us the sort of contradiction most scoring apps ship with — "Mostly fine. Many processed ingredients." — where the grade said one thing and the description said the opposite.
Unifying the engine around one question made every grade in the app consistent with every other number we show. If the nutrition looks good and the grade looks bad, there's no contradiction — those are measuring different things.
We also made the call to never add positive offsets to the grade. The grade only goes down. "A" isn't a reward for being healthy — it's what you get when there's nothing much to flag. Healthiness is a nutrition conversation.
What we publish, what we don't. This post describes what goes into the grade and why. We keep the exact coefficients for each risk tier, the numeric boundaries between letters, our full functional-ingredient exclusion list, and our combination rules in the app itself. Publishing them would invite reformulation games — tweak a recipe to land just under a threshold. The philosophy is public; the dials aren't.
We'll update the rubric when the science does. If EFSA reclassifies an additive, or if NOVA 4 markers shift, the app updates. We'll say publicly when we move a big line.
You scan a barcode. In under two seconds you get a letter, a plain-English verdict, and a separate nutrition tab if you want to dig deeper. The letter is an answer to one narrow question. The rest of the screen is how we widen it without muddling the grade.
Scanning, grading and the full E-number breakdown all sit inside NutraSafe on iOS.
Get NutraSafe on the App Store