Scoring methodology · Drinks
Every drink is evaluated against 6 rules, worth a total of 115 points. Your score is the percentage of points the product earns on what we actually know about it.
Quality score
Each rule contributes points up to a known maximum. We sum them, divide by the total possible, and that's the headline number on the product.
Unknowns
If we couldn't verify an attribute, that rule earns zero. Brands aren't rewarded for non-disclosure, and we don't fill the gap with a guess.
clean_ingredients
No harmful contaminants flagged. Severity- and amount-weighted against per-serving food guidelines (MADL) using the shared food-path penalty math.
lab_indexed
A full third-party lab report has been indexed for this product. Beverages in particular benefit from independent verification because contaminant leaching (PFAS, BPA, heavy metals) varies widely by packaging and supply chain.
packaging
Container material. Glass jars and paperboard score at the top; PET / polystyrene / generic plastic at the bottom. Generic "container" / "bottle" with no material signal lands mid-tier.
recognized_certifications
Product carries at least one recognized third-party certification (USDA Organic, Non-GMO Project, Regenerative Organic, Fair Trade, etc.). Binary credit — verification of any one counts.
added_sugar
Per-serving added sugar with drink-realistic thresholds (most sodas are 30–40 g per serving). ≤2 g earns full 15 (unsweetened or very light); 2–8 g partial; 8–16 g reduced; 16–30 g minimal; >30 g zero. Falls back to "Total Sugars" when "Added Sugars" is undisclosed (less precise — naturally-occurring sugars count too).
artificial_sweeteners
Penalizes presence of aspartame, sucralose, acesulfame-K, saccharin, neotame, and advantame in the ingredient list. Each detected sweetener nibbles 4 points; first detected sweetener brings the rule down from 10 to 6. Allulose, stevia, and monk fruit are deliberately NOT on this list — they're natural-derived non-nutritive sweeteners with cleaner profiles. No ingredient list = unknown / partial credit.
Scoring is a function applied to evidence. If you think a rule fired on the wrong evidence, the per-product breakdown shows the source for every contribution — each one is challengeable. The methodology itself is version-controlled and evolves in public.
