← Back to search

About Evidence Audit

Twins Skin · Research transparency tool

What this tool does

Evidence Audit lets anyone search peer-reviewed skincare literature and instantly see who funded each study. Every result carries an Evidence Score — a deterministic, 0–100 composite of study design, funding independence, sample size, recency, and citation count. The goal is one sentence: make it harder to cite industry-funded research as if it were independent truth.

How studies are audited

When you submit a query, the backend fans out to three public bibliographic APIs — PubMed, OpenAlex, and EuropePMC — and merges the results. Each study then passes through a two-pass classifier: a fast deterministic pass (exact string matching against a curated brand conflict registry) followed by an LLM pass (GPT-4o-mini) that reads the acknowledgements and funding disclosure fields and assigns one of four funding tags: independent, mixed, unclear, or industry.

Results are cached by query hash for 24 hours. Cache hits are served in under 100 ms; cold searches take 3–8 seconds depending on API latency. No personally identifiable data is stored; only the normalised query string and a hashed IP are logged for rate-limit and abuse purposes.

Evidence Score formula

The score is a deterministic sum of six dimensions, capped at 100. A higher score does not mean a study is "true" — it means it is methodologically stronger and less likely to be biased by funding.

DimensionMax ptsHow it is scored
Design40Meta-analysis 40 · Systematic review 38 · RCT 35 · Observational 20 · Review 15 · Case series 10 · Other 5
Independence25Independent 25 · Mixed 12 · Unclear 8 · Industry 0
Size15Log₂ scaled, saturates at n = 500 participants
Recency10Full 10 pts if ≤ 5 years old; linearly decays to 0 at 15 years
Citations5Saturates at 100 citations
Bonus5Awarded when query contains a comparison term (e.g. "vs", "versus", "compared to")
Total100Capped at 100

Brand conflict registry

The deterministic pass checks funding disclosures against the following registered parent companies and their known brands and aliases. Studies funded by any of these entities receive an industry tag and score 0 on the Independence dimension.

Parent companyBrands / aliases
AmorepacificLaneige, Sulwhasoo, Innisfree, Etude
Bayer ConsumerBayer, Coppertone, Bepanthen
BeiersdorfBeiersdorf AG, Nivea, Eucerin, Aquaphor, La Prairie
Estée LauderEstee Lauder Companies, Clinique, La Mer, Origins, Bobbi Brown, MAC, Aveda
GaldermaCetaphil, Differin, Restylane
Johnson & JohnsonJ&J, Band-Aid, Aveeno Baby
KaoKao Corporation, Bioré, Curel, Molton Brown, John Frieda
KenvueJohnson & Johnson Consumer, Neutrogena, Aveeno, Clean & Clear, OGX
L'OréalL'Oreal, Loreal, CeraVe, La Roche-Posay, Vichy, SkinCeuticals, Garnier, Lancôme, Kiehl's
LVMH BeautyLVMH, Fresh, Benefit, Guerlain, Dior Beauty
P&GProcter & Gamble, Procter and Gamble, Olay, SK-II, Pantene
Pierre FabrePierre Fabre Dermo-Cosmetique, Avène, A-Derma, Ducray, Klorane, René Furterer
ShiseidoShiseido, Clé de Peau, NARS, Drunk Elephant
UnileverDove, Vaseline, Pond's, Murad, Dermalogica, Paula's Choice
Walgreens Boots AllianceWBA, No7, Soap & Glory, Liz Earle, Boots

Why transparency matters

Studies funded by the companies whose products they test are more likely to report favourable outcomes — a well-documented phenomenon across medicine and nutrition science. Skincare is no different. At Twins Skin we sourced and audited our product recommendations against primary literature, and we noticed how often the most-cited papers were funded by the brands making the claims. This tool exists so anyone — not just researchers — can apply the same scrutiny in seconds. We do not hide industry-funded studies; we surface them with their funding status clearly labelled so you can weigh the evidence yourself.

Contact

Questions, corrections, or requests to add sponsors to the registry: [email protected]