Methodology
How we score
Every Tallow score is a single number from 0 to 100. It is not a vibe and it is not a black box: a product starts at 100 and earns specific, published penalties for what is actually in it and how it is made. The same math runs on every product, so two people who scan the same item always see the same score — and you can audit every line of it in the app on the “Why this score” breakdown.
How the number is built
Every product begins at a perfect 100. Penalties come off from there.
Every ingredient carries a severity rating from 0 (worst) to 4 (beneficial). The worst tier costs 40 points, the next 10, mild-caution ingredients 3, and neutral or beneficial ingredients cost nothing. Penalties stack — a mild ingredient can never dilute a terrible one.
Ingredients are listed by quantity, so a bad ingredient at the top of the list hurts more: the first ingredient's penalty is multiplied by 1.3, easing down to ×1.0 for the last.
The combined ingredient penalty is capped at 85 points. Two worst-tier ingredients already hit the cap — a genuinely harmful formula lands deep in Poor no matter what else is on the label.
A contaminant an independent lab actually measured is scored against its published limit rather than by severity rating: a reading at the limit costs the full 50 points, half the limit costs 25, and the readings sum to a cap of 50. A substance a lab looked for and did not detect costs nothing at all. A substance nobody has measured is not treated as absent — it keeps its ordinary ingredient penalty.
Industrial formulations built from additives, isolates, and ingredients you would not find in a home kitchen.
Processed culinary products — not whole foods, but not ultra-processed either.
Packaging is rated 0 (worst) to 4 (best) by material: the riskiest food-contact packaging costs 8 points, the best costs nothing. Unknown packaging is never penalized.
The only bonus in the whole model. Nothing else buys points back.
The total is rounded once and clamped to the 0–100 range.
One honest edge case: a product with no ingredient list at all is shown as Unrated — never a misleading 100. No data is not the same as clean.
The ingredient severity ladder
Every ingredient in our database is rated 0 (worst) to 4 (beneficial). The penalty ladder is deliberately steep — a worst-tier ingredient costs four times a tier-1 and thirteen times a tier-2 — because one truly harmful ingredient should matter more than a pile of mild flags.
The worst offenders. One alone forces a Poor score.
Clearly harmful, but a notch below the worst tier.
Mild flags. A couple of these still leave a product Clean.
Neither good nor bad — costs nothing, buys nothing back.
Genuinely good ingredients. Highlighted, but they never add points — good stuff can't paper over bad stuff.
Score bands
Poor
0–40
Okay
41–69
Clean
70–100
What is NOVA?
NOVA is a food-classification system from researchers at the University of São Paulo, widely used in nutrition science. It groups foods by how much industrial processing they have been through — which is why it carries weight in the score.
- NOVA 1UnprocessedWhole foods — fresh fruit, eggs, plain milk, raw meat.
- NOVA 2Culinary ingredientsPressed, ground, or refined from NOVA 1 — olive oil, butter, salt.
- NOVA 3ProcessedNOVA 1 or 2 combined and preserved — canned beans, fresh bread, simple cheese.
- NOVA 4Ultra-processedIndustrial formulations with additives and isolates — soda, packaged snacks, most cereals.
Kid-Safe Mode
Kid-Safe Mode does exactly one thing: it subtracts a flat 5 points from any product that has never been lab tested at all. Every other term — ingredients, contaminants, processing, packaging, organic — is identical in both modes, so the Kid-Safe number is always just the ordinary score minus 5, or the ordinary score unchanged when a lab has tested the product.
It used to multiply every penalty by 1.5. That was removed: scaling the penalties while leaving the organic bonus flat quietly handicapped organic products, and the mode as a whole was hard to reason about. Feeding a child is the case where an unverified label deserves the least benefit of the doubt, so that is the only thing the toggle now expresses.
Your standards
Alongside the number, Tallow can give a personal thumbs-up or thumbs-down against the standards you pick. It is one-strike: if any standard you selected fails, the product is rejected for you — regardless of its score. When the data needed for a check is missing, that check is marked unknown rather than quietly passed, and if every check is unknown you see “not enough info” instead of a fake approval. The standards you can pick:
- Artificial ingredientsAny red-tier ingredient — dyes, synthetic sweeteners, harsh preservatives.
- Ultra-processedThe product is NOVA group 4.
- Seed oilsCanola, soybean, corn, sunflower & co. — including the label names that hide them, like “vegetable oil”.
- High sugarOver 22.5g per 100g (11.25g for drinks — the FSA “high” thresholds).
- High carbsOver 30g of carbohydrates per 100g.
- Risky packagingThe two worst packaging tiers — plastic and risky food contact.
- Hidden contaminantsThe product must have been lab tested, and every measured contaminant must have come in under 25% of its published limit. This is the one standard where missing data fails rather than reading unknown — an untested product fails it, because “nobody looked” is the thing this check exists to catch.
Restaurants
The restaurant map applies the same lens to places to eat — seed oils, additives, plastic contact, and premade food — but the scores are built from community ratings, weighted by the quality of each reviewer’s evidence. That model has its own page: how we score restaurants.
No black boxes, no paid placements. We will never accept payment from the brands or restaurants we rate.