Turn the promise into a concrete question
“Improves skin” is too broad. Ask which measure changed: hydration, barrier function, perceived comfort or something else? On which area, after how long and with which directions? A concrete question quickly reveals whether the promise has substance.
Then match the product. Did the research examine the same formula, concentration, packaging, group and method of use? If not, the source may explain an ingredient, but it does not establish the result of the bottle in your hand.
Different sources answer different questions
A laboratory experiment can explore a mechanism or signal. An ingredient study tells you what happened with a specific material. A controlled trial of the finished product gets closer to the question a buyer cares about: what happened under defined conditions of use.
A testimonial can be a useful account of one person's experience, but it cannot establish cause. Even a strong study has boundaries: a forearm result does not automatically apply to the face, and evidence in reactive skin does not cover every user.
Two real examples of a conclusion you should not transfer
In a 1999 study, a specific sonicated S. thermophilus preparation was applied to the forearms of 17 healthy volunteers for seven days, and an increase in stratum-corneum ceramides was observed. This does not demonstrate a benefit for every cosmetic described as probiotic.
In another study, 66 women with reactive skin used a cream containing 10% B. longum lysate for two months. The reported changes belong to that lysate, concentration and formula — not the whole category and not ART BEAUTY products.
Green and red signals in beauty copy
A green signal is a specific outcome, the name of the finished product, a clear method, duration and group. A direct source link and visible limitations are reassuring. The EU common criteria for cosmetic claims emphasise truthfulness, evidential support and honesty.
A red signal is “scientists proved” without a link, an ingredient result assigned to the whole formula, medical language for cosmetic care or a percentage without explaining what was measured. Public copy can be concise, but never stronger than the data.
SAVE BEFORE YOU BUY
Your quick checklist
- What is promised and how was it measured?
- Was the ingredient or the finished product studied?
- Do the site, dose and duration match?
- Was there a control or meaningful comparison?
- Can you open the primary source?
Sources
Direct links to the research and documents used in this article. They correspond to the numbered citations above.
- Commission Regulation (EU) No 655/2013: common criteria for cosmetic claims.
- Di Marzio L et al. Topical sonicated Streptococcus thermophilus and stratum-corneum ceramides. Journal of Investigative Dermatology, 1999.
- Guéniche A et al. A Bifidobacterium longum lysate cream in reactive skin. Experimental Dermatology, 2010.