Methodology

How Luminelle scores.

A Luminelle score is a regulatory-scrutiny score, not a safety verdict. It answers one question: how much of what is on this label do regulators restrict, and how much of that restriction can be read off the label itself? It says nothing about whether a product works, and it is not medical advice.

This page is the whole method. The same reasons and the same points appear against every ingredient inside the app, so anything you read here can be checked on any product you scan.

The short version

  • Every ingredient is matched to the regulators' own entries, and each entry is read for what it actually says about this kind of product.
  • Each ingredient lands in exactly one of three tiers: a finding, something under scrutiny, or listed but not scored.
  • A product starts at 100 and loses points per tier. Repeated small flags saturate, so ten preservatives do not read like a ban.
  • The label's own order is used as evidence. EU law orders ingredients by weight only above 1%, and Luminelle uses that line to prove that capped ingredients sit within their cap.
  • Products that stay on the skin are scored harder than products that are rinsed off.

Why not simply flag every restricted ingredient

Most scanners score presence: if an ingredient appears on any list, the product loses points. That makes mainstream, fully compliant products score badly on preservatives and UV filters the EU explicitly permits, and it treats a limit on how much may be used as if it were a ban. The criticism levelled at this approach is fair. A limit is not a hazard, and a label cannot show a concentration.

Luminelle's answer is to score what the entry means rather than the fact that an entry exists. A ban is a finding whatever the dose. A concentration limit is scrutiny, because the label cannot prove the level used. And a limit the label can prove, through ingredient order, is not scored at all.

Tier one: findings

Dose-independent. These are the only reasons a product can be called out for something banned, and each costs the points shown, per ingredient.

Reason shown in the appPoints
Banned in cosmetics (EU Annex II, or a California ban already in force).20
Banned, with an exception this label gives no reason to apply.20
Banned for this kind of product.20
The EU chemicals agency concluded it is an endocrine disruptor.20
Permitted only in other kinds of product.12

Tier two: scrutiny

Real restrictions whose weight depends on something the label cannot show: the concentration used, or an assessment that has not concluded. Classes marked as saturating follow a curve rather than a straight sum, so the first flag of a kind costs the most and the tenth costs almost nothing. For a product that stays on the skin, every scrutiny total is multiplied by 1.4.

Reason shown in the appPoints
Banned when used as a fragrance ingredient.10
A ban has been passed but is not yet in force.10
Under assessment by the EU chemicals agency as a possible endocrine disruptor.8
Listed under California Proposition 65.6
Permitted up to a limit for this kind of product. The label cannot show the level used.4, saturating at 30
Restricted in the EU, with no limit text on the register entry.3, saturating at 8
Permitted up to a limit, and listed after ingredients that are usually under 1%.2, saturating at 15
A fragrance allergen, named on the label because EU law requires it.2, saturating at 14

The saturating curve is cap × (1 − e−n/k), where n is the number of ingredients in the class and k is 4 for concentration limits, 5 for fragrance allergens and 2 for other restrictions. Two capped ingredients cost about 12 points; six cost about 23; the total can never pass 30.

The leave-on multiplier is deliberately timid. The EU's Scientific Committee on Consumer Safety assumes roughly a hundredfold difference in retention between a product left on the skin and one rinsed off. Luminelle applies 1.4.

Tier three: listed, not scored

Shown against the ingredient so that nothing is hidden, but worth zero points, because the entry does not apply to this product as labelled.

  • Permitted up to a limit, and listed after the 1% line, so the label itself shows it is within the limit.
  • Regulated for this kind of product, but the condition is not a limit on how much can be used.
  • The ban is on a contaminated grade of the material. Cosmetic grade is the grade that passes.
  • The ban is on the nano form. A nano ingredient must say [nano] on the label, and this one does not.
  • The ban is on material of human origin. Cosmetic supply is animal or plant.
  • The register itself says the entry applies only under a condition the label cannot show.
  • A purity rule on the raw material, not a limit on the product.
  • Restricted for a specific use, such as hair dye, that this product is not.
  • Assessed by the EU chemicals agency and concluded not to be an endocrine disruptor.
  • A permitted colourant on the EU list, with no limit attached.
  • Listed by an industrial or disclosure database only. Not a cosmetic restriction.

The score

Each ingredient counts once, in its most severe class. The class totals are added, the scrutiny totals are scaled for leave-on products, and the result is subtracted from 100. The score is clamped to 0 to 100. The per-ingredient points shown in the app are whole numbers that add up to exactly the penalty, so the ledger always reconciles with the score.

The 1% line

Article 19 of Regulation 1223/2009 requires ingredients to be listed in descending order of weight, but only down to 1%. Below that, any order is allowed. Luminelle turns this into evidence. The first ingredient whose legal limit for this kind of product is 1% or lower marks the start of the sub-1% zone. Anything capped at 1% or more that appears after that marker is proven within its limit by the label alone, and is moved to the not-scored tier.

A softer line uses ingredients that are conventionally formulated under 1%, such as EDTA salts, gums, pH adjusters, tocopherol and the named fragrance allergens. A cap proven only against one of these is not a legal fact, so it halves the penalty instead of removing it.

Product type

Many register entries carry different limits for rinse-off, leave-on, oral, nail, deodorant, hair, lip, eye or hand products. Luminelle infers the product type from its category and name, then reads the sub-entry that applies. An entry written for a use this product is not, such as hair dye, is not scored. If a product's type cannot be told, it is treated as leave-on, which is the stricter reading. An entry that is ambiguous for a type is never counted as a violation.

A worked example

CeraVe Foaming Cleanser for Normal to Oily Skin, 236 ml (EAN 3337875597197). Eighteen ingredients resolve to regulator entries. Four carry a flag.

  1. Sodium benzoate, eighth on the label. EU Annex V/1 permits it in rinse-off products up to 2.5%. The label cannot show the level used, so it is scrutiny, under a concentration limit.
  2. Phenoxyethanol, twelfth. Annex V/29 permits it up to 1%. This is the first ingredient capped at 1% or lower, so it marks the 1% line. It is itself under a concentration limit.
  3. Tetrasodium EDTA and ethylhexylglycerin appear only on ECHA's industrial classification inventory. Listed, not scored.

Two ingredients in the concentration-limit class: 30 × (1 − e−2/4) is 11.8 points. A cleanser is rinsed off, so no multiplier. The score is 100 − 11.8, which rounds to 88. Nothing banned, two things under scrutiny, two things listed for transparency.

Sources

Every flag has a primary source, and every ingredient page in the app links to the original entry. These are the lists, what each one does in the score, and how often Luminelle refreshes it.

SourceRoleRefresh
EU CosIng (Annexes II to VI)Bans, restrictions, concentration limits, product-type conditions, permitted colourants and preservatives.Monthly
Regulation (EC) 1223/2009The legal text behind CosIng, including Article 19 labelling rules.Quarterly
ECHA endocrine disruptor assessment listConcluded and pending assessments. Only a concluded endocrine disruptor is a finding.Quarterly
SCCS opinionsScientific Committee on Consumer Safety opinions, cited on ingredient pages. Not scored.Monthly
ECHA classification and labelling inventoryIndustrial hazard classifications. Shown for transparency, never scored.Quarterly
California AB 2762 and AB 2771Cosmetic ingredient bans in force. Scored as bans.Fixed text
California AB 496Bans effective 31 December 2026. Scored as upcoming until then.Fixed text
California OEHHA Proposition 65Chemicals known to the state to cause cancer or reproductive harm.Monthly
California CDPH Safe Cosmetics ProgramReported ingredients and the reportable ingredient list. Shown, not scored.Weekly and monthly
US EPA PFAS master listIdentifies PFAS for the sensitivity filter. Not scored.Quarterly

What the score does not do

It does not estimate exposure, and it does not know the concentration used unless the label proves it. It does not score ingredient quality, efficacy, texture, price or ethics. It does not use hazard classifications written for factories and transport. And it never says an ingredient is toxic or dangerous. The words used are the regulators' own: banned, restricted, permitted up to a limit, under assessment.

Corrections

If you think an ingredient is classified wrongly, or an entry is missing, write to info@luminelle.ai with the regulator citation. Corrections are applied at the next source refresh.

Scoring methodology: tiers, points and sources | Luminelle