Trust by design

How Filtizy scoring works

A score should help you ask better questions, not hide the details. Every product page shows the exact factors, raw values, and weights that produced its number - nothing is a black box.

Food & Beverages · Health & Wellness

Scored per 100g / 100ml of the product

Calories12%
Protein13%
Carbohydrates5%
Total sugar7%
Added sugar15%
Fibre13%
Total fat5%
Saturated fat10%
Sodium5%
Ingredient quality10%
Processing5%

Each percentage is that factor's share of this track's weighting - added sugar counts three times as much as sodium, for example. Protein and fibre reward higher values; calories, carbohydrates, both sugars, both fats and sodium all reward lower ones. Nine of the eleven read one number straight off the pack's nutrition panel; the last two - Ingredient quality and Processing - are read from the ingredient list instead, counting flagged additives and how many ingredients are named. Each factor is bucketed into one of ten score tiers (100 down to 10, in steps of 10) against the thresholds below. A product whose panel is entirely blank scores 0, not 100: an empty label is missing data, not a perfect one.

Every other category

Electronics, fashion, home, beauty and more

Quality40%
Value30%
Data trust20%
Availability10%

Each percentage is that factor's share of this track's weighting. These categories have no natural per-100g unit to score against, so an admin assesses each factor directly on a 0-100 scale at listing time instead of a range ladder.

The formula

Filtizy score = the weighted average of every factor that applies to a product's category, rounded to the nearest whole number:

Σ (factor score × factor weight) ÷ Σ (factor weight)

A product is only ever scored against one track - a nutrition-scored product is never judged on quality/value/trust/availability, and vice versa. Dividing by the sum of that track's own weights means every score is always out of 100% of the factors that actually apply to it, regardless of how many factors that track has.

Nutrition scoring thresholds

Panel bands are per 100g; Ingredient quality counts flagged additives and Processing counts listed ingredients, so neither is a per-100g figure. Each factor's ten rungs are placed on the decile boundaries of our own catalogue, so every rung separates real products rather than sitting empty; where a published band exists it is pinned to a rung so the top of the scale keeps an absolute meaning - UK FSA traffic-light green/red for sugar, fat, saturated fat and salt, and the EU "source of / high in" claim bands for protein and fibre.

Factor	100	90	80	70	60	50	40	30	20	10
Calories	≤ 150kcal	≤ 300kcal	≤ 370kcal	≤ 385kcal	≤ 415kcal	≤ 455kcal	≤ 475kcal	≤ 510kcal	≤ 570kcal	above 570kcal
Protein	≥ 26g	≥ 20g	≥ 12g	≥ 10g	≥ 8.2g	≥ 7g	≥ 6.4g	≥ 5.9g	≥ 5g	below 5g
Carbohydrates	≤ 10g	≤ 20g	≤ 30g	≤ 47g	≤ 56g	≤ 65g	≤ 69g	≤ 73g	≤ 80g	above 80g
Total sugar	≤ 3g	≤ 5g	≤ 8g	≤ 10g	≤ 13.5g	≤ 19g	≤ 22.5g	≤ 28g	≤ 33g	above 33g
Added sugar	≤ 1g	≤ 3g	≤ 5g	≤ 6.6g	≤ 8.2g	≤ 11g	≤ 19g	≤ 24g	≤ 31g	above 31g
Fibre	≥ 11.8g	≥ 10g	≥ 8.5g	≥ 7.8g	≥ 6.3g	≥ 6g	≥ 5.4g	≥ 4.5g	≥ 3g	below 3g
Total fat	≤ 1.5g	≤ 3g	≤ 6g	≤ 10.6g	≤ 12.7g	≤ 17.5g	≤ 19.5g	≤ 23.8g	≤ 35g	above 35g
Saturated fat	≤ 0.7g	≤ 1.5g	≤ 2.4g	≤ 4.1g	≤ 5g	≤ 6g	≤ 8g	≤ 9.7g	≤ 12g	above 12g
Sodium	≤ 42mg	≤ 90mg	≤ 120mg	≤ 152mg	≤ 202mg	≤ 250mg	≤ 285mg	≤ 350mg	≤ 430mg	above 430mg
Ingredient quality	≤ 0 flags	≤ 1 flags	≤ 2 flags	≤ 3 flags	≤ 4 flags	≤ 5 flags	≤ 6 flags	≤ 7 flags	≤ 8 flags	above 8 flags
Processing	≤ 3 ingredients	≤ 5 ingredients	≤ 7 ingredients	≤ 10 ingredients	≤ 12 ingredients	≤ 13 ingredients	≤ 14 ingredients	≤ 16 ingredients	≤ 18 ingredients	above 18 ingredients

Why each factor is scored this way

Every direction below was checked against established food-labeling methodology rather than assumed - Nutri-Score (the EU/France official algorithm), UK FSA traffic-light labeling, and NOVA (the WHO-referenced ultra-processed food classification).

CaloriesLower is better
Energy density is one of Nutri-Score's four official negative components, alongside sugar, saturated fat and sodium. It's a blunt measure, though: it can't tell why a food is calorie-dense, so nutrient-rich but energy-dense foods (nuts, oils) can score worse here than they deserve - Protein and Fibre exist specifically to balance that out. The top rung is held at 150kcal rather than at our catalogue's first decile of 368kcal, because 370kcal is typical for a packaged shelf but is not, in absolute terms, a low-calorie food.
ProteinHigher is better
An explicit positive component in Nutri-Score - higher protein supports satiety and muscle maintenance. 5g ("source of protein") and 10g ("high in") are the EU claim bands; the rungs above them separate protein bars and powders, which cluster at 19-32g, from biscuits at 6-8g.
CarbohydratesLower is better
A total-carbohydrate read rather than a "low carb" claim: the rungs walk from a vegetable or dairy base through a grain or flour base to confectionery. Weighted below the sugar factors on purpose - carbohydrate on its own says much less about a product than how much of it is sugar. The upper rungs sit close together because that is where our catalogue actually is, with a median of 65g.
Total sugarLower is better
Matches UK FSA traffic-light bands (green at 5g, red above 22.5g per 100g), which are pinned to the 90 and 40 rungs; the remaining rungs follow the distribution of our own catalogue.
Added sugarLower is better
Weighted heaviest of any nutrition factor because it's the more evidence-backed metric: WHO's guideline specifically targets free/added sugar (under 10%, ideally under 5% of energy intake), since that evidence is stronger than for total sugar, which also includes sugar naturally present in dairy and whole fruit. The top rung is strict - 1g or less is effectively "no added sugar", and a declared 0 is a real value that earns it.
FibreHigher is better
Also an explicit Nutri-Score positive component - supports digestion, blood sugar control and satiety. 3g ("source of fibre") and 6g ("high in") are the EU claim bands; above them the rungs separate mueslis and bran cereals.
Total fatLower is better
UK FSA front-of-pack total fat bands (green at 3g, red above 17.5g per 100g). Weighted below saturated fat because total fat lumps together fats with very different effects - the saturated share is the part the evidence singles out. Nut butters land in the bottom rung, which is the right read for a fat factor even though they score well on protein and fibre.
Saturated fatLower is better
Matches UK FSA traffic-light bands (green at 1.5g, red above 5g per 100g), pinned to the 90 and 60 rungs.
SodiumLower is better
UK FSA's salt thresholds converted to sodium (salt is ~39% sodium by mass) put green at ~120mg, which is the 80 rung. The rungs below it are tight around 150-290mg because that is where most of the shelf sits - our catalogue's median is 218mg.
Ingredient qualityLower is better
The one factor read from the ingredient list rather than the nutrition panel. It counts distinct additive concerns - artificial colours and flavours, preservatives, emulsifiers, stabilisers, artificial sweeteners, hydrogenated oil, high-fructose corn syrup, MSG and E-number additives - which are the same categories NOVA identifies as ultra-processing markers. Each distinct concern costs a rung. Note the known hole: a product that publishes no ingredient list at all scores 100 here, ahead of one that discloses honestly.
ProcessingLower is better
Also read from the ingredient list: how many ingredients are named. Fewer scores higher, following the common "5-7 ingredient" rule of thumb for spotting minimally processed food. Weighted lightest of the two ingredient-based factors because research shows ingredient count alone is a weaker signal than ingredient type - Ingredient quality above captures that stronger signal. The middle rungs step one item at a time because our catalogue's median is 12 ingredients, so a single item is a real difference.

This is a prototype heuristic (v0.2) for application development, not a scientifically validated methodology. Before making public health or quality claims, validate the thresholds with qualified domain expertise and publish the methodology transparently.

Why version scores?

Every score is stored alongside the factor rows that produced it - nothing is recomputed on the fly. If the formula improves later, existing products keep their historical breakdown and can be rescored deliberately, with a clear audit trail rather than scores silently drifting.

See it on real products →