Which ingredient pairs were ever tested together

Section 01

What we actually searched

We took 20 of the ingredients people ask about most and looked at every pairing between them that gets recommended online — 59 pairs in all. For each one we searched the public trial registry for a single intervention naming both ingredients: one product, or one regimen given inside one arm of a study. That is the strict reading of "tested together", and it is the only reading that tells you whether anyone measured the combination rather than the parts.[1][2]

We counted a looser category separately. When a study names both ingredients but in different arms, they were compared against each other, not combined. That answers a different question — which one wins — and it says nothing about what happens when you put both on your face on the same evening.[1]

Section 02

Six pairs out of fifty-nine

Fifty-three of the 59 pairs have no completed skin trial with posted results. Nineteen have never been named together inside a single intervention at all, in any trial, finished or running. The remaining pairs sit in between: a trial exists, but it has not finished, or it finished and never posted what it found.[1][2]

The distribution is lopsided. Adapalene with benzoyl peroxide has 42 skin trials naming both in one intervention, 37 completed, 27 with posted results. Hydroquinone with tretinoin has 11, only one of which posted results. Benzoyl peroxide with tretinoin has 6. After that the numbers fall to three, two, one, zero.[1]

That shape has an obvious cause. Combinations get properly tested when a company wants to sell them as one licensed product and needs a regulator to agree. Nobody funds a trial of two serums you were going to buy separately anyway.[3][2]

Section 03

The pairings people ask about most

Niacinamide with vitamin C is the clearest example of the gap between noise and evidence. It has been argued about for years, and the registry holds two skin trials naming both in one intervention, neither with posted results. The argument is louder than the data underneath it.[1][5]

Section 04

What this changes for your shelf

It removes the false authority from layering rules. When someone tells you two ingredients cancel each other out, ask what measured that. For most pairs the honest answer is that nothing did, and the rule came from a plausible-sounding argument about pH or oxidation rather than from anyone putting both on skin and counting the result.[1][5]

It also cuts the other way. An untested pair is not endorsed. If you combine two strong actives and your skin stings, peels or reddens, that is real information, and it arrived faster than any trial would have. Drop back to one active, let the skin settle, and reintroduce slowly.[4][5]

The practical shortcut: use tested combinations where they exist and you need the result. Adapalene with benzoyl peroxide for acne is the one pairing on this list where the evidence is genuinely deep, and it exists as a single product so you are not guessing at the ratio.[3][4]

Section 05

Where this count is blind

Registry intervention fields are free text written by sponsors. A trial that used a combination but described it only inside an attached protocol will not appear. Cosmetic products are also under-represented, because a cosmetic company can sell without registering a trial at all, so the absence of skincare pairs partly reflects who is required to publish.[2][1]

Published research outside the registry exists too, especially older split-face work. Our search is a floor, not a ceiling. What it does establish is that no large, modern, results-posted body of evidence sits behind the layering rules people repeat as fact.[5][2]