DemythSkin is a free skincare ingredient checker that scores both efficacy and safety. Type a product name or scan a barcode; the engine retrieves the full INCI list, classifies every ingredient by clinical evidence, and returns two scores out of 5.0 plus a plain-language verdict. The same rubric applies to a $9 drugstore cleanser and a $400 luxury cream — the formula is all that's scored.

Those tools answer one question: “is this safe?” DemythSkin also answers the one you're actually asking: “will this work?” A product can be perfectly safe and completely useless — safety-only scoring can't see the difference. Our efficacy score checks whether the actives have clinical evidence, whether they're present at meaningful concentrations, and whether the formulation supports them.

Both are out of 5.0. Efficacy measures how likely the formula is to deliver its claims — driven by evidence-graded actives, INCI position (a proxy for concentration), and ingredient synergies, with penalties for “fairy-dusted” actives listed at trace amounts. Safety measures irritation potential, with deductions weighted by severity: documented sensitizers like added fragrance and drying alcohol count heavily; rare, low-frequency sensitizers barely register. Full math on the methodology page.

Every ingredient lands in one of five tiers by clinical evidence: S — Gold Standards (retinoids, L-ascorbic acid, proven UV filters), A — Reliable (ceramides, niacinamide, glycerin, panthenol), B — Supporting (soothing botanicals, emollients, developing evidence), C — Overhyped | Marketing (heavily hyped, thin clinical support), D — Documented Irritants (fragrance, drying alcohol, established sensitizers). Preservatives, thickeners, and other formulation-base ingredients are shown separately — they make a product stable, not effective, so they're neither rewarded nor punished.

No — and that's deliberate. Standard preservatives like phenoxyethanol are what keep your products from growing bacteria; treating them as hazards is fearmongering, not science. DemythSkin's D Tier is reserved for ingredients with documented sensitization evidence or regulatory restrictions. Ingredients with genuinely rare reaction reports get a small “rare sensitizer” note — visible if you're reactive and looking, never inflated into a warning.

Yes. When a formula scores well (or badly), the analyzer surfaces alternatives with equal or better ingredient profiles — matched by mechanism and concern, not just category — and ingredient-level dupes with a formula-overlap percentage. Independence applies here too: recommendations follow the scores, never a deal.

Extensively. The database covers hundreds of K-beauty brands — COSRX, Beauty of Joseon, SKIN1004, Anua, Round Lab, Torriden and more — scored on the same evidence rubric as Western brands. Ferments, centella, and snail mucin get graded on their actual clinical support, which is sometimes better and sometimes worse than the hype suggests.

The analyzer is free, no account needed. A free account unlocks alternatives & dupes, and lets you build a skin profile. Pro then applies that profile to every scan — flagging ingredients on your avoid list, malassezia triggers if you're fungal-acne prone, and conflicts with products you already own. Nothing about the scores themselves is ever paid or paywalled by brands: Efficacy and Safety are identical for subscribers and non-subscribers.

Six questions — skin type, sensitivity, main concern, pregnancy, whether you're fungal-acne prone, and any ingredients you want flagged. It takes about a minute and costs nothing. Every scan then answers a third question beside the two scores: is this right for you? A gold-standard product can still be wrong for one person, and that's exactly what a profile catches. Set one up.

Both, with a strict division of labor. AI retrieves formulas and writes the plain-language explanations. The scores come from a deterministic rubric — ingredient tiers, position weighting, severity-graded safety deductions — anchored in published research. AI explains; the rubric decides. Details on the methodology page.