- Establishes
- A marketing tier. Sometimes used by clinics to mean 'sold through a practice,' not a regulated class.
- Does not establish
- It does not mean prescription-only, FDA-approved as a drug, or more effective than a drugstore formula with the same active and dose.
- Ask instead
- What does the full ingredient list say, at what percentage, and in what base? Who actually made it?
- Source
- No FDA, EMA or ISO class is named 'medical grade' for cosmetics. A clinic channel is not a regulation.
Claim literacy
What a term establishes.
What it does not.
No toxin scores. No fear marketing. Product-type logic over prestige.
Paste a sentence from the box
A match is not evidence. The decoder names the claim and where it sits on the scale. It will not clear a product for pregnancy, a procedure, or a prescription.
The evidence scale
Every claim on a box sits somewhere on this scale. Knowing which step is most of the work; the rest is deciding whether that step is enough for the decision in front of you.
- 01Established
Repeated human studies measuring the thing you actually care about — SPF testing, or the well-studied ways retinoids are used.
- 02Supported
There is human data, but it is small, tied to one particular base, or borrowed from a close relative of this ingredient.
- 03Mixed
Early clinical work, mixed results, or a match on the product name alone. Useful as a hunch, not a verdict.
- 04Mechanistic only
We can describe how it might work. That is not the same as it working on a face over 12 weeks.
- 05Thin
A word that sells a feeling. It does not bind a dose, a test, or a timeline.
- 06No adequate evidence
The word is claiming more than the public evidence supports. Rather than guess, we name what is missing.
Terms, decoded
- Establishes
- That a dermatologist was involved in some test. How the test was run is almost never printed on the box.
- Does not establish
- It does not mean hypoallergenic, non-comedogenic, or safe for a damaged barrier. How many people were tested, and what was measured, are usually left unsaid.
- Ask instead
- How many people, measuring what, on which finished formula — and was it a 24-hour patch test or eight weeks of real use?
- Source
- A process claim. Without n, endpoint, duration and the finished formula, a test is not a result.
- Establishes
- A test occurred in a clinical-adjacent setting. That is a process claim, not an outcome.
- Does not establish
- It does not mean clinically proven, peer-reviewed, or better than vehicle. 'Tested' is not 'worked.'
- Ask instead
- Ask what it was compared against, how many people took part, how long it ran, and whether the test used the formula you can actually buy.
- Source
- 'Tested' names a setting. It does not name a comparison, a population, or a change that held.
Clinically proven
No adequate evidence
Long version in the publication ↗Decode it on a treatment menu ↗
- Establishes
- Nothing, until the study is named. “Proven” is a verdict word; it needs a study behind it.
- Does not establish
- It does not establish magnitude, duration, or relevance to your skin job. A 4-week shine study is not 12-week pigment evidence.
- Ask instead
- Name the study, what it measured, and whether it was run on this product or a similar one from the same range.
- Source
- A verdict without a named study, population, endpoint and formula cannot be checked. The desk will not invent one.
- Establishes
- A hope that common sensitizers were reduced. The word is not a legal standard in most markets.
- Does not establish
- It does not mean you will not react. Fragrance-free is a more checkable claim than hypoallergenic.
- Ask instead
- Is it fragrance-free? Free of essential oils? And how many people were in the repeat-patch allergy test?
- Source
- US and most EU cosmetic rules do not define 'hypoallergenic.' It is a shopper word, not a standard.
- Establishes
- At best, a rabbit-ear or human-back comedogenicity assay on a formula — often unpublished.
- Does not establish
- It does not mean the product will not congest your face. Occlusion, climate, and how much you use still matter.
- Ask instead
- Where was it tested, on what skin, and is the current formula the one that was tested?
- Source
- Comedogenicity assays, when they exist, rarely match real-world dose, climate or a damaged barrier.
- Establishes
- A retail philosophy. The excluded list changes by retailer.
- Does not establish
- It does not mean safer, more effective, or fragrance-free. 'Clean' is not a toxicology result.
- Ask instead
- Which ingredients were excluded, and what replaced them? Replacement is the real formula.
- Source
- Clean is a shelf tag. Sephora's list is not Credo's, and neither is a toxicology review.
- Establishes
- A mood. Some botanicals are well-studied; many are not.
- Does not establish
- Natural is not gentler. Poison ivy is natural. Essential oils are a common irritation source.
- Ask instead
- Name the ingredient, the dose, and the evidence for that dose — not the farm.
- Source
- Origin is not a safety or efficacy standard. Poison ivy is natural.
- Establishes
- A real job: slow the water your skin loses, restore comfort, replace the fats in the barrier. Ceramide, cholesterol and fatty-acid blends have genuine research behind them.
- Does not establish
- It does not mean a three-day transformation, a microbiome reset, or permission to keep using a stripping cleanser.
- Ask instead
- Does the formula contain a ceramide:cholesterol:fatty-acid logic, and will you stop the thing that is breaking the barrier?
- Source
- Occlusion and lipid replacement are plausible and studied. A 7-day 'repair' claim on a serum is not that literature.
- Establishes
- A research field. Consumer products rarely publish strain-level, durability, or clinical endpoints.
- Does not establish
- It does not mean the product will recolonize your face or outperform a bland cleanser plus moisturizer.
- Ask instead
- Is a specific strain named, is it still alive at the end of shelf life, and was anything measured on human skin — or does the carton just say 'prebiotic'?
- Source
- Strain, viability at end of shelf life, and a human endpoint are the evidence. 'Prebiotic' on a carton is not.
- Establishes
- A desired look: radiance or a more even tone. Could mean pigment, glow, or just more light scatter from glycerin. Some brightening formulas do contain actives with research behind them, such as niacinamide, vitamin C or azelaic acid.
- Does not establish
- It is not a regulated efficacy term and does not name the mechanism (melanin, keratin, optical). Glow from film is not pigment change, and the word does not show a product fades dark marks or melasma. Illegal products containing mercury or hydroquinone have been sold as 'brightening,' so the word tells you nothing about what is in the jar.
- Ask instead
- Which active, at what dose, is there a full ingredient list, and are you wearing enough SPF to make the test fair?
- Source
- Brightness is not a pigment diagnosis. Without the molecule, the dose and the wear, the word is atmosphere.
Whitening / lightening / bleaching
Thin
Long version in the publication ↗Decode it on a treatment menu ↗
- Establishes
- That the product is meant to reduce overall skin pigment. Lighter skin is not a health goal, and this desk does not treat it as one.
- Does not establish
- It does not show the product treats a medical pigment condition, or that it contains only what the label says. In the US, over-the-counter skin-lightening drug products are not legally marketed. Products sold under these words have been found with mercury, hydroquinone or hidden steroids.
- Ask instead
- Is there a full ingredient list at all? Does it name mercury, mercurio, calomel or mercurous chloride, or list hydroquinone without a prescription? If the real job is a dark mark, what caused the mark?
- Source
- Unlabeled bleaching products still contain mercury or high-dose steroids. No INCI, no desk rating.
Dark spot corrector
No adequate evidence
Long version in the publication ↗Decode it on a treatment menu ↗
- Establishes
- An intent to target local dark marks rather than overall tone. That is a job description, not a result.
- Does not establish
- It does not show the product works, how fast, or whether it suits marks left by breakouts versus melasma. It does not show it will not irritate, and irritation can leave new marks. A US over-the-counter corrector listing hydroquinone is not legally marketed.
- Ask instead
- Which active, for which kind of mark, over how many months, and did everyone in the study wear daily sunscreen?
- Source
- A mark is not one disease. PIH, melasma and a lentigo do not share a corrector.
- Establishes
- A marketing phrase. At most, a hint that the brand designed for, or tested on, deeper skin tones.
- Does not establish
- No regulatory definition was found. It does not show low irritation or a low chance of dark marks for your skin, and no word on a box can. Procedures described this way still carry a higher chance of dark marks, lighter patches and burns on deeper skin.
- Ask instead
- Tested on how many people with deeper skin tones, over how long, measuring what? For a procedure: how often does this provider treat skin like yours, and will they do a test spot first?
- Source
- No regulator defines the phrase. Inclusive copy is not a panel, and a procedure word does not lower risk.
- Establishes
- An intention that the product is for broad use. On sunscreens and tinted products it is usually about how it looks on skin.
- Does not establish
- It does not show the product was tested on lighter, medium and deeper skin tones alike, that it leaves no white cast, or that it protects against visible light.
- Ask instead
- Tested on which skin tones, with how many people in each, and was the result how it looked, how it felt, or how it protected?
- Source
- Inclusive copy is not a panel. Ask who was photographed, and how change was read without relying on pink.
- Establishes
- That hydroquinone is not in the formula.
- Does not establish
- It does not show the replacement works as well or irritates less. Some alternatives, such as cysteamine, have trial data; many do not. 'Free-from' names what is missing, not what replaced it.
- Ask instead
- What replaced it, at what percentage, and is there a study on that active for the kind of mark you have?
- Source
- 'Free-from' names an absence. The replacement, the dose and the mark type are the evidence.
- Establishes
- A hope that common irritants were left out. Sensitive skin is something people report about their own skin, and it is reported across every background.
- Does not establish
- It is not a legal standard and does not mean you will not react. It does not mean fragrance-free, and no skin tone or heritage is more sensitive by default.
- Ask instead
- Is it fragrance-free and free of essential oils, which actives does it contain, and was it tested on people who describe their own skin as sensitive?
- Source
- Sensitive is not a diagnosis and not a patch-test result. The INCI is the evidence.
- Establishes
- A category aisle. Aging is not a single pathway a cream can reverse.
- Does not establish
- It does not specify photoaging vs volume vs texture vs pigment. Most of the honest work is SPF + retinoid-class + time.
- Ask instead
- Which visible change, over what timeline, compared against what — and was daily sunscreen part of the study?
- Source
- Aging is not one endpoint. Population-level evidence sits with UV protection, then retinoids as a class.
- Establishes
- A class of signaling fragments. Some have in-vitro and small clinical signals; most on a label are unnamed and undosed.
- Does not establish
- Peptide is not a dose. 'Peptide complex' without a percentage is a guest list, not a recipe.
- Ask instead
- Which peptide, what percentage, in what base, and was the study run on this formula?
- Source
- In-vitro signaling is not a finished-formula trial. Name the peptide and the human data, or it stays a hunch.
- Establishes
- A family of proteins/peptides that can signal in tissue. Source, concentration, and stability are the whole story.
- Does not establish
- It does not mean the product contains human growth hormone, nor that a cream can replicate a procedure.
- Ask instead
- Where it came from (human, plant, or lab-made), at what concentration, and whether this is a cosmetic or something only a clinic should apply.
- Source
- Wound-healing protein papers do not transfer to a 30 ml serum. Name the source, the dose and the finished-formula data.
- Establishes
- Usually plant stem-cell culture media or lysates. A plant cell cannot become your skin cell.
- Does not establish
- It does not mean live human stem cells are in the jar. That would be a different regulatory object.
- Ask instead
- Is it a plant lysate, a conditioned media, or an actual cell therapy? Those are not interchangeable.
- Source
- A plant culture lysate is not a cell therapy. The words on the box are usually the former.
Exosome terminology
No adequate evidence
Long version in the publication ↗Decode it on a treatment menu ↗
- Establishes
- A research object: extracellular vesicles. Clinic add-ons often do not disclose source, isolation, or contents.
- Does not establish
- A named buzzword is not a characterized product. Source and sterility are unresolved until written.
- Ask instead
- Species/source, isolation method, protein/RNA characterization, and the regulatory pathway in your jurisdiction.
- Source
- Wound-healing vesicle papers do not transfer to a cosmetic add-on. Without source, isolation and contents, unresolved.
- Establishes
- Nothing physiological in a cosmetic context. The liver does not need a clay mask.
- Does not establish
- It does not mean impurities are extracted from tissue in a way that changes health.
- Ask instead
- If the job is oil or debris on the surface, say cleanse. If the job is inflammation, name the active.
- Source
- No cosmetic 'detox' endpoint exists in human physiology. The liver is not a clay mask.
- Establishes
- Temporary optical tightening from astringents or films. Pore openings are anatomical.
- Does not establish
- It does not mean pores close or disappear. Congestion can look smaller when keratin is managed.
- Ask instead
- Is the job oil, keratin, or camera light? Treat the job; do not buy a vanishing act.
- Source
- Pore diameter is anatomy. A toner cannot resize it. Oil, light and swelling change how it photographs.
- Establishes
- Words like 'sun protection' or 'UV defense' without a tested SPF number. Atmosphere, not a method.
- Does not establish
- 'Antioxidant shield,' 'pollution defense,' or 'UV-friendly botanicals' are not SPF. A tint that sheers to a stain is not a dose. And SPF and Broad Spectrum say nothing about visible light.
- Ask instead
- What is the SPF number on this unit, the filters, and will you apply enough film — about 1/4 teaspoon for a face?
- Source
- SPF is a lab method on a finished formula. 'UV defense' on a moisturizer without a number is not that method.
- Establishes
- A tested UVB (and usually UVA) protection factor on the finished formula, at a stated application thickness.
- Does not establish
- It does not mean you applied that thickness, reapplied, or covered visible light. A number on the box is not the film you wore.
- Ask instead
- What is the number, is it broad-spectrum, and how much did you actually put on — about 1/4 teaspoon for a face?
- Source
- SPF is an FDA monograph / ISO 24444 lab method. The number is the rating. Wear and reapplication are the intervention.
- Establishes
- An appearance claim: how the product looks once it is on.
- Does not establish
- It does not show how it looks on every deeper skin tone, how much UVA protection it gives, or anything about visible light. Untinted 'invisible' formulas generally do not block visible light.
- Ask instead
- Seen on which skin tones, in what light, at the full amount you would actually apply? If visible light matters for your marks, is there a tinted version?
- Source
- Cast is a camera and undertone problem. Invisible often means under-applied, or photographed on one phototype.
Blue light / HEV protection
No adequate evidence
Long version in the publication ↗Decode it on a treatment menu ↗
- Establishes
- Some intent to address visible light, the part of sunlight that SPF does not measure.
- Does not establish
- No standard test or label measure for it was found. The claim does not say how much is blocked, or whether this formula blocks any. It is not SPF and does not replace UV protection.
- Ask instead
- How was it measured, against what, and what is doing the work: a tint from iron oxides, or an antioxidant in the ingredient list?
- Source
- Visible light is a pigment trigger. Iron oxides in a tint are the practical filter; a badge on an untinted cream is not.
- Establishes
- Tinted formulas colored with iron oxides can block visible light, which can darken pigment in medium and deeper skin. In melasma studies, UV plus visible-light protection did better than UV protection alone.
- Does not establish
- No standard label figure says how much visible light is blocked. It does not promise a shade that matches your skin, and a tint is not a UV dose: the SPF number still has to be there.
- Ask instead
- Are iron oxides in the ingredient list, is there a shade that matches you, and does the same unit carry an SPF number and Broad Spectrum?
- Source
- Melasma trials of UV plus visible-light protection used tinted films with iron oxides. The tint is the filter; the SPF number is still required.
Retinoid / retinol / adapalene
Supported
Long version in the publication ↗Decode it on a treatment menu ↗
- Establishes
- A class with human data for acne and photoaging. Adapalene 0.1% is the congestion lead most people actually keep.
- Does not establish
- It does not mean this percentage in this vehicle was the trial, or that nightly use is tolerated on day one.
- Ask instead
- Which retinoid, what strength, how many nights, and is SPF the next morning part of the product?
- Source
- Class evidence for retinoids is strong. A named bottle is usually unpublished. Frequency, not a new brand, is the variable.
- Establishes
- That no fragrance was added as a scent. A real reduction in common sensitizers when it is true.
- Does not establish
- It does not mean unscented masking agents are absent, or that essential oils were not used as 'botanicals.'
- Ask instead
- Is it fragrance-free on the INCI, or just unscented? Any essential oils?
- Source
- Fragrance-free is a useful filter. It is not a patch test, and 'unscented' can still mask.
The same reading, pointed at a treatment
Everything above is about what a sentence on a bottle establishes. The identical problem exists one aisle over, in the language a clinic or spa uses to describe a procedure — and it is usually worse there, because the sums are larger and the claims are harder to check. Its sibling desk handles that side.
Continue in Spa Intelligence · Claim Decoder
Decode a treatment claim the same way
The same sentence-level reading, pointed at a spa or clinic advertisement instead of a bottle. Paste one line of their marketing and it separates what was actually named from what was conspicuously left out.
Carries nothing you have to protect — you paste the sentence there. Your routine stays in this browser.
Continue in Spa Intelligence · Reference library
Look up what a treatment class has to disclose
“Medical grade” means as little on a treatment menu as it does on a bottle. The library sets out what each service class is actually obliged to name, and where a credential can be checked.
Carries nothing. It is a reference, not a form.
Things we could not find a source for
These are all widely repeated, including by people who ought to know. We went looking and came back without an adequate source. That is recorded here rather than quietly left out, because an unmarked popular claim is how this desk would start misleading people.
The claim
Copper peptides firm skin or rival established anti-aging actives.
What we found
No systematic review or guideline was found supporting topical copper peptides. The one meta-analysis of peptides for skin aging is dominated by oral collagen peptides — only two of nineteen trials studied a topical formulation, and copper peptides do not appear in it at all.
The claim
'Purging' lasts four to six weeks, happens only where you normally break out, and can be told apart from a reaction by how it looks.
What we found
No trial, review, guideline or consensus statement defining purging or validating a timeframe was found. What does exist: the tretinoin label describes an apparent early increase in inflammatory lesions, and the adapalene label documents that irritation peaks in the first four weeks and usually settles. Guideline bodies set the review point at 12 weeks. The specific popular claims have no support we could find.
The claim
Hyaluronic acid draws water out of your skin in dry air unless you seal it.
What we found
Stated in tertiary textbook sources about humectants as a class, but cited to no experiment. The most directly relevant measurement points the other way: at 30% relative humidity a hyaluronic acid fluid decreased water loss rather than increasing it. Layering over damp skin and sealing is sound formulation practice; the warning behind it is not established.
The claim
Mineral sunscreen filters are the ones to choose in pregnancy.
What we found
No obstetric or dermatology body was found making a mineral-versus-chemical recommendation for pregnancy. Regulators have established that some organic filters are absorbed, and that absorption alone does not establish harm. Preferring mineral filters is a reasonable personal choice; skipping sunscreen over the question is not.
The claim
Ceramide moisturizers outperform other moisturizers.
What we found
The Cochrane evidence is for moisturizers as a class. One 2025 intra-participant trial in eczema-prone adults found physiological lipids reduced water loss where a glycerine-only emollient did not — promising, single-center, four weeks, not definitive.
References
Everything this desk cites, with the date each was last read. A citation is not an endorsement — several are here because they contradict a claim rather than support one.
- What to look for: ABCDEs of melanoma (opens in a new tab)
Clinical guideline
American Academy of Dermatology, 2026
Asymmetry, irregular border, varied color, diameter, and evolution as the features that should prompt an appointment — as should any spot that is new, different from the others, or changing, itching or bleeding.
Read 2026-09-02.
- Acne vulgaris: management (NG198) (opens in a new tab)
Clinical guideline
National Institute for Health and Care Excellence, 2021
Fixed-combination topical treatments — adapalene with benzoyl peroxide, or tretinoin with clindamycin — as first-line 12-week options at any acne severity, reviewed at 12 weeks. Sets referral criteria for nodulo-cystic acne, scarring, and acne causing psychological distress.
Does not establish: That any single topical outperforms the others, or that results should be judged before 12 weeks.
Read 2026-09-02.
- 21 CFR 201.327 — Over-the-counter sunscreen labeling and effectiveness testing (opens in a new tab)
Regulatory / official
US Code of Federal Regulations, 2024
The only skin-cancer claim permitted for broad-spectrum SPF 15+ is that it decreases risk *if used as directed with other sun protection measures*. Also the source of the 2 mg/cm² figure: that is the dose at which SPF is measured in the lab, in the test procedure itself. Reapplication is mandated at least every two hours, and after swimming, sweating or towel drying.
Does not establish: That 2 mg/cm² is a clinically derived recommendation. It is a laboratory test condition, which is why applying less does not simply scale the number down.
Read 2026-09-02.
- Your skin and the sun (opens in a new tab)
Regulatory / official
Oxford University Hospitals NHS Foundation Trust, 2024
Studies show people frequently apply less than half the sunscreen needed for the protection on the label. Practical rule given as at least six teaspoons for the body.
Does not establish: A single agreed quantity — the WHO's 2025 review gives about nine teaspoons for the same job. The disagreement is real and worth stating.
Read 2026-09-02.
- Atopic eczema (opens in a new tab)
Regulatory / official
NHS, 2024
Urgent advice if skin is blistered, crusty, leaking fluid or has pus-filled spots; is painful, swollen or warm; suddenly gets worse; or comes with a high temperature or feeling generally unwell. These can indicate infection or eczema herpeticum.
Read 2026-09-02.
- Cellulitis (opens in a new tab)
Regulatory / official
NHS, 2024
Skin that is painful, hot and swollen needs urgent attention. Emergency care for a very high temperature, fast heartbeat or breathing, purple patches, confusion, or cold clammy skin — these are sepsis signs.
Read 2026-09-02.
- FDA works to protect consumers from potentially harmful OTC skin lightening products (opens in a new tab)
Regulatory / official
US Food and Drug Administration (Drug Safety Communication), 2022
Under the CARES Act, over-the-counter skin lighteners containing hydroquinone needed FDA approval to be legally marketed from 23 September 2020, and there are no approved OTC skin lighteners. Reported harms include rash, facial swelling and exogenous ochronosis, which may be permanent. At the date of the notice (April 2022) the only FDA-approved hydroquinone product was a prescription triple-combination cream for melasma.
Does not establish: That hydroquinone does not work — the finding is regulatory, not about efficacy. It says nothing about prescription products after April 2022, or about any market outside the US: the EU cosmetics ban this entry used to mention is not on this page.
Read 2026-09-13.
- Mercury Poisoning Linked to Skin Products (opens in a new tab)
Regulatory / official
US Food and Drug Administration, Consumer Updates, 2022
Mercury is illegally added to some skin lighteners and anti-aging creams. Label terms to watch: mercurous chloride, calomel, mercuric, mercurio, mercury. These products are often sold in shops serving Latino, Asian, African and Middle Eastern communities, and online. Health effects are listed, and if exposed FDA's steps are to stop use, wash, contact poison control, and seal the product.
Does not establish: That every unlabeled product contains mercury, or how common exposure is.
Read 2026-09-13.
- Alpha Hydroxy Acids (opens in a new tab)
Regulatory / official
US Food and Drug Administration, Cosmetic Ingredients, 2022
FDA's own studies found four weeks of AHA use produced an 18% increase in UV-induced erythema and roughly doubled sensitivity to UV cell damage, reversible on stopping. Hence the recommended sunburn alert covering use and the week afterwards.
Does not establish: Wrinkle or collagen claims at cosmetic concentrations. The sun-sensitivity finding is the better-established fact about AHAs.
Read 2026-09-02.
- Hypoallergenic Cosmetics (opens in a new tab)
Regulatory / official
US Food and Drug Administration, 2022
There are no federal standards or definitions governing the term, and manufacturers are not required to submit substantiation to the FDA.
Read 2026-09-02.
- RENOVA (tretinoin cream) 0.02% prescribing information (opens in a new tab)
Regulatory / official
US Food and Drug Administration, NDA 021108, 2014
Adjunctive mitigation of fine facial wrinkles, alongside a comprehensive skin-care and sun-avoidance program.
Does not establish: The label states in capitals that it does not eliminate wrinkles, repair sun-damaged skin, reverse photoaging or restore younger skin, and that it has not shown effect on coarse wrinkling, roughness, mottled pigmentation, lentigines, telangiectasia or laxity.
Read 2026-09-02.
- Questions and Answers: FDA announces new requirements for OTC sunscreen products (opens in a new tab)
Regulatory / official
US Food and Drug Administration, 2011
SPF measures protection against UVB (sunburn). 'Broad Spectrum' means UVA protection in proportion to UVB protection.
Does not establish: Anything about visible light, which it does not address at all. The label rules in force today may have been updated since 2011; that was not checked.
Read 2026-09-13.
- Visible light. Part II: Photoprotection against visible and ultraviolet light (opens in a new tab)
Clinical review
J Am Acad Dermatol, 2021
Organic and inorganic UV filters do not protect against visible light; tinted sunscreens do. Regulation of visible-light protection differs by region.
Does not establish: A standard for testing or labeling visible-light protection — it does not define one, so it cannot say how much a given tinted product blocks.
Read 2026-09-13.
- Topical, light-based and complementary interventions for acne: an overview of systematic reviews (opens in a new tab)
Systematic review
Cochrane Database of Systematic Reviews, 2024
The largest current synthesis of acne topicals — 275 trials, 40,910 participants. Its headline finding is that there is no high-certainty evidence for the effects of any therapy included.
Does not establish: Any confident ranking of acne topicals against each other. This is the reason every acne rating on this desk is hedged: strong guideline recommendations sit on a low-certainty trial base.
Read 2026-09-02.
- Niacinamide: a review on dermal delivery strategies and clinical evidence (opens in a new tab)
Systematic review
Drug Delivery and Translational Research 14(12):3512–3548, 2024
Pigment is niacinamide's strongest domain: 4% performed comparably to 4% hydroquinone in melasma, and 1–4% over 8–12 weeks reduced facial hyperpigmentation.
Does not establish: The oil-control claim. The review found essentially no published randomized trials measuring sebum — that claim rests on a 2002 conference presentation. Barrier claims are largely mechanistic.
Read 2026-09-02.
- Tranexamic acid as a therapeutic option for melasma management: meta-analysis of RCTs (opens in a new tab)
Systematic review
J Dermatolog Treat, 2024
22 randomized trials, 1,280 patients: tranexamic acid reduced melasma severity. The largest MASI drop was with oral use, then injection, then topical. Side effects included stomach upset and menstrual changes.
Does not establish: Long-term safety, which is not established. Heterogeneity between trials was high.
Read 2026-09-13.
- Efficacy and safety of cysteamine 5% cream for melasma: systematic review and meta-analysis of RCTs (opens in a new tab)
Systematic review
Arch Dermatol Res, 2024
7 randomized trials: cysteamine beat placebo, with no significant difference from hydroquinone 4%. Irritation rates were similar to hydroquinone and higher than placebo.
Does not establish: That cysteamine and hydroquinone are equivalent — few, small trials cannot show that.
Read 2026-09-13.
- Self-applied topical interventions for melasma: a systematic review and meta-analysis of randomized, investigator-blinded trials (opens in a new tab)
Systematic review
British Journal of Dermatology 187(3):309–317, 2022
36 randomized trials. Topical tranexamic acid 2.5% against vehicle reduced MASI at high certainty — the only high-certainty rating in the review. Hydroquinone 4% beats placebo; triple-combination cream beats hydroquinone alone but with markedly more adverse effects.
Does not establish: A meaningful difference between hydroquinone 4% and tranexamic acid on MASI — that comparison found none.
Read 2026-09-02.
- Topical benzoyl peroxide for acne (opens in a new tab)
Systematic review
Cochrane Database of Systematic Reviews, 2020
120 trials, 29,592 participants. Benzoyl peroxide may improve self-reported acne compared with placebo (low certainty), with little to no difference against adapalene or clindamycin.
Does not establish: Superiority over a retinoid, or good tolerability — withdrawal for adverse effects was roughly doubled against placebo.
Read 2026-09-02.
- Topical azelaic acid, salicylic acid, nicotinamide, sulphur, zinc and fruit acid (alpha-hydroxy acid) for acne (opens in a new tab)
Systematic review
Cochrane Database of Systematic Reviews, 2020
Salicylic acid shows little to no difference against tretinoin on treatment response (low certainty). Azelaic acid probably performs worse than benzoyl peroxide for acne (moderate certainty).
Does not establish: Any claim that salicylic acid is a leading acne active. Risk of bias and imprecision limit confidence throughout.
Read 2026-09-02.
- Interventions for rosacea based on the phenotype approach: an updated systematic review with GRADE assessments (opens in a new tab)
Systematic review
British Journal of Dermatology 181(1):65–79, 2019
152 trials, 20,944 participants. High-certainty evidence for topical azelaic acid and ivermectin for the papules and pustules of rosacea.
Does not establish: Azelaic acid for rosacea redness or flushing — that evidence belongs to brimonidine and oxymetazoline, not azelaic acid.
Read 2026-09-02.
- Emollients and moisturisers for eczema (opens in a new tab)
Systematic review
Cochrane Database of Systematic Reviews, 2017
77 studies, 6,603 participants. Moisturizers reduce eczema flares and topical steroid use, and work better alongside an active treatment than the active treatment alone.
Does not establish: That ceramide-containing moisturizers outperform other moisturizers — the benefit shown is for moisturizers as a class.
Read 2026-09-02. Published 2017 and not since updated.
Journal of Cosmetic Science 68(4):255–256, 2017
There is a hierarchy of comedogenicity test methods but no mandated one. An in-use clinical trial with dermatologist grading is described as the gold standard; the rabbit-ear model is the least rigorous of the recognized options.
Does not establish: That 'non-comedogenic' on a label guarantees any testing happened, or that per-ingredient comedogenicity ratings predict whether a finished formula will break out a given person.
Read 2026-09-02.
- Systematic review of randomized controlled trials on interventions for melasma: an abridged Cochrane review (opens in a new tab)
Systematic review
J Am Acad Dermatol, 2014
20 randomized trials, 2,125 participants: triple-combination cream beat hydroquinone alone (RR 1.58), azelaic acid 20% beat hydroquinone 2% (RR 1.25), and tretinoin beat placebo.
Does not establish: Long-term outcomes or relapse. Methodology was poor and the studies were short.
Read 2026-09-13.
British Journal of Dermatology 180(2):289–296, 2019
44 participants over 12 weeks. Both bakuchiol and retinol reduced wrinkle area and hyperpigmentation, with no statistically significant difference between them; retinol users reported more scaling and stinging.
Does not establish: Equivalence. There was no placebo arm, the dosing was unbalanced (twice daily against once daily), and 44 participants cannot establish non-inferiority — 'no significant difference' in a small trial is not evidence of sameness. The tolerability finding is the more credible half.
Read 2026-09-02.
Photodermatol Photoimmunol Photomed, 2014
68 melasma patients, all also using 4% hydroquinone for 8 weeks: a UV-plus-visible-light (iron oxide) sunscreen gave greater improvement than a UV-only sunscreen on MASI, colorimetry and melanin.
Does not establish: What a tinted sunscreen does on its own — everyone also used hydroquinone. Single-center and short-term, and it tells us nothing about dark marks after inflammation (PIH).
Read 2026-09-13.
Annals of Internal Medicine 158:781–790, 2013
903 adults under 55 in Queensland. Daily SPF15 against discretionary use produced 24% less skin aging by microtopography over 4.5 years (odds 0.76, 95% CI 0.59–0.98).
Does not establish: Generalization beyond a high-UV setting and a mostly fair-skinned cohort using SPF15.
Read 2026-09-02.
Journal of Clinical Oncology 29(3):257–263, 2011
1,621 participants followed to 2006: invasive melanoma hazard ratio 0.27 (95% CI 0.08–0.97).
Does not establish: A significant reduction in all melanoma — that result (HR 0.50, p=0.051) did not reach conventional significance.
Read 2026-09-02.
J Invest Dermatol 130(8), 2010
Controlled experimental study in 20 volunteers with skin types IV–VI: visible light produced darker and more sustained pigmentation than UVA1. There was no pigmentation in type II skin.
Does not establish: Anything about everyday sunlight or sunscreen. It used lab light sources on the lower back and measured no sunscreen effect.
Read 2026-09-13.
The guide library
24 short decision frameworks. The desk quotes these at you when they become relevant; this is the whole set, readable on its own terms.
Where this goes next
Reading a claim well is only useful attached to a decision. These are the decisions this desk can actually help with.
Put it in an order
A claim about an ingredient means very little until it has a position in a sequence and a frequency.
Check what you already own
Most of the answer to “should I buy this” is already on your shelf, filed under a different claim.
Nothing is working
If the claims all check out and the routine still is not moving, the problem is usually order, dose or time — not the label.