Nearly every skincare brand cites studies. “Clinically proven.” “Backed by research.” “In a study, 90% of users saw improvement.” The phrases are so common that they have stopped meaning much, which is a shame, because the difference between a real study and a marketing prop is knowable — and you do not need a statistics degree to spot it. You need five questions.
Here they are, in the order worth asking them.
1. Was there a control — and was it the right one?
A study without a comparison group tells you almost nothing. If everyone in the study used the product and skin improved, you cannot say the product did it. Skin changes with the weather, the season, and the simple act of paying attention to it. The comparison is what isolates the ingredient.
The comparison that matters most in skincare is the vehicle control. The vehicle is the base formula minus the active ingredient — the same cream, the same serum, without the thing being tested. A good study compares the active formula against its own vehicle, because a moisturizer with no active in it still improves skin. If a niacinamide serum only beats “no treatment,” some of that win is just moisturizer. If it beats its own vehicle, the niacinamide is doing something. When a study brags about results “versus baseline” with no vehicle arm, that is the first place to be skeptical.
2. Who — or what — was actually studied?
This is the single biggest gap between what a label implies and what a study shows. There is an enormous difference between an ingredient tested on human skin and one tested on cells in a dish or on animals.
Cell-culture and animal studies are real science, and they are how ingredients get discovered. But they answer “can this compound do something under ideal conditions,” not “will this product do something on my face.” A compound that quiets inflammatory signaling in a dish of macrophages is promising. It is not the same as a proven result on skin, and a brand that quotes the dish study as if it were the face study is counting on you not to ask.
This is exactly the line our own evidence tiers are built on: an ingredient with human, controlled trials sits higher than one supported mainly by lab-model work, and we say which is which out loud. We wrote up how that grading works in what our evidence tiers actually mean.
3. How many people, and for how long?
Sample size and duration are the least glamorous numbers and two of the most important. A result from ten people over two weeks is a hint. A result from a hundred people over twelve weeks is evidence. Neither is worthless, but they do not deserve the same confidence, and marketing loves to quote the small, short study in the same certain tone as the large one.
Duration matters especially in skincare, because most real changes are slow. Cell turnover, the fading of a dark spot, the softening of a line — these unfold over weeks to months. A two-week study measuring a twelve-week outcome is measuring the wrong window, and an impressive short-term number is sometimes just the hydration bump any decent formula gives on day one.
4. Who paid for it — and did they publish the misses?
Industry funding does not automatically invalidate a study; much good cosmetic research is funded by industry, because who else would fund it. But it is a reason to read more carefully, not less. The question underneath the funding question is publication bias: are you seeing the study because it worked, while the ones that did not work quietly never appeared? A single glowing result, with no independent replication and no disclosed funding, is the weakest kind of citation, however confident the sentence around it sounds.
5. What did it actually measure?
Finally, check whether the study measured the thing you care about or a stand-in for it — a surrogate endpoint. “Increased collagen production in laboratory-cultured cells” sounds like “fewer wrinkles,” but it is a lab measurement, several steps removed from anything visible in a mirror. “Improved skin barrier markers” is not the same as “skin looked and felt better.” Surrogate endpoints are legitimate in research and slippery in marketing, because they let a brand imply a visible result that the study never actually showed.
A worked example
Put the five questions to a real ingredient. Niacinamide — the active in our Clarity Serum — has been studied in split-face human trials, where one side of a person's face gets the active and the other gets the vehicle, over the kind of timeline that matters, measuring the appearance outcomes people actually care about. Control, human subjects, reasonable numbers and duration, real endpoints. That is why niacinamide earns a top grade on our scale, and why we are comfortable putting it front and center.
Contrast that with an ingredient whose case rests mainly on a cell-culture study showing an effect in a dish. It may well be a good ingredient. But it has not cleared the same bar, and honesty means grading it lower and saying so — which is what “Mixed” means when you see it on one of our pages. We would rather stamp Mixed and cite what exists than borrow certainty a study did not earn.
The habit
You do not have to run these questions like a checklist forever. After a while it becomes a reflex — you see “clinically proven,” and instead of feeling reassured, you get curious. Proven on whom? Against what? For how long? Measuring what? Most marketing claims cannot survive four of those questions, let alone five, and learning to ask them is the most durable skincare skill there is. It outlasts any single product, including ours.
If you would rather skip straight to a routine built on ingredients that clear the bar, the Skin Plan maps yours in about thirty seconds — no email required. And the full research library is where we show our work.
References
- Bissett DL, et al. Niacinamide: a B vitamin that improves aging facial skin appearance. Int J Cosmet Sci. 2004. pubmed.ncbi.nlm.nih.gov/18492135
- Kim S, et al. Inhibitory effects of (−)-α-bisabolol on LPS-induced inflammatory response in RAW264.7 macrophages — an example of cell-culture evidence. Food Chem Toxicol. 2011. sciencedirect.com
- Sorrel & Co. What our evidence tiers actually mean. sorrel.skin/blogs/research