Skin measurement with an error bar
Most skin trackers report your lighting as progress. Assay measures its own error first, then calls a change real only when it beats that margin on your face, on your device. It is the only skin tracker that tells you when it cannot tell.
$170
spent a year trying to find skincare that is right for you
4
failed products the average woman is still holding on to
1 in 9
women who have cycled through ten or more failures
£1bn
of skincare abandoned annually in the UK alone
First three: AmLactin / OnePoll survey of 2,000 American women aged 35 and over, February 2022. Fourth: The Body Shop survey of 2,000 UK adults, reported by Professional Beauty. The same OnePoll survey found a product gets about three weeks to prove itself, which is shorter than the onset window of almost every active ingredient it might contain.
Nobody can tell you which four. The measurement exists, YouCam’s Skin Analysis API scores sixteen skin attributes from a photograph, but a single score carries no error bar, and without one you cannot separate a working product from a brighter bathroom.
This is not an accusation, it is in the documentation. Perfect Corp describe the displayed value as “a psychological motivator”, adjusted upward from the underlying measurement because “consumers generally prefer positive evaluations regarding their skin health.”
It is a reasonable product decision. It is also why this exists. I measured the gap on my own face before I read the reason for it, and it is largest on moisture, the one concern a hydrating product is supposed to move. Assay computes everything on the measurement and never on the motivator.
68
displayed
45.4
measured
A 22.6 point upward adjustment. A tracker scoring progress on the left-hand number reads an encouragement as evidence.
These are the calibration sessions, captured 15 minutes apart. Skin cannot change in 15 minutes, so every difference below is measurement error, and there is nothing real here for anything to find.
A tracker that shows you a score and compares it to last time would have reported 10 of these 12 comparisons as a change in your skin, the largest of them +17.0 points of redness. Every one would have been false.
Assay reported none of them. Not because it is cautious, but because it had measured what this instrument does when nothing happens, and all of it fell inside that.
| Concern | Sessions | Difference | A tracker says |
|---|---|---|---|
| Redness | 1 → 2 | -4.18 | changed |
| Redness | 2 → 3 | +17.04 | changed |
| Pores | 1 → 2 | -0.18 | no change |
| Pores | 2 → 3 | +1.68 | changed |
| Texture | 1 → 2 | +1.36 | changed |
| Texture | 2 → 3 | -2.22 | changed |
| Radiance | 1 → 2 | -3.45 | changed |
| Radiance | 2 → 3 | -6.09 | changed |
| Blemishes | 1 → 2 | +3.68 | changed |
| Blemishes | 2 → 3 | -4.23 | changed |
| Hydration | 1 → 2 | +0.14 | no change |
| Hydration | 2 → 3 | -1.85 | changed |
10 false claims from a tracker without an error bar. 2 honest ones, by luck rather than method. Assay: 0 claims, all 12 correctly held below the floor.
A prospective n-of-1 study running on a real face, with real products, scored by the YouCam Skin Analysis API. Every reading is captured live. Nothing on this page is simulated.
Surface smoothness and roughness.
Change
-7.5
Noise floor
±3.1
Texture moved 7.5 points in the wrong direction, past the 3.1-point noise floor. That is outside the window where this product is expected to cause temporary irritation. Worth stopping and reconsidering.
Active spots and inflammatory lesions.
Change
-6.9
Noise floor
±6.4
Blemishes moved 6.9 points in the wrong direction, past the 6.4-point noise floor. That is outside the window where this product is expected to cause temporary irritation. Worth stopping and reconsidering.
Visible flushing and irritation. The first thing to move when something disagrees with you.
Change
+7.1
Noise floor
±24.6
Redness has moved 7.1 points, inside the 24.6-point margin of error. That is not yet distinguishable from measurement noise, and at the current noise level a 5-point change is not resolvable on a practical schedule. Improving capture consistency would lower the floor.
How much water the surface layer of your skin is holding.
Change
-2.1
Noise floor
±2.9
Hydration has moved 2.1 points, inside the 2.9-point margin of error. That is not yet distinguishable from measurement noise, and at the current noise level a 5-point change is not resolvable on a practical schedule. Improving capture consistency would lower the floor.
A tracker without a noise floor would have read this as -6.9 and told you it was declining.
Visible pore size and density.
Change
-2.2
Noise floor
±4.0
Pores has moved 2.2 points, inside the 4-point margin of error. That is not yet distinguishable from measurement noise. 11 more sessions, about 11 days, would let us resolve a 5-point change.
How evenly your skin reflects light. Highly sensitive to the light you photograph it in.
Change
-12.5
Noise floor
±13.4
Radiance has moved 12.5 points, inside the 13.4-point margin of error. That is not yet distinguishable from measurement noise. 12 more sessions, about 12 days, would let us resolve a 5-point change.
3 calibration sessions sampled from separate windows of one continuous 15-minute recording. Same person, same seat, same camera, no product. Nothing about this skin changed, so everything below is the instrument, not the face.
| Concern | Readings, 15 min apart | Vs. within-sitting | Drift | Noise floor | Can it answer? |
|---|---|---|---|---|---|
| Hydration | 87.8 88.0 86.1 | -1.7 | ±2.9 | Measurable | |
| Texture | 96.3 97.6 95.4 | -0.9 | ±3.1 | Measurable | |
| Pores | 92.9 92.8 94.4 | +1.5 | ±4.0 | Marginal | |
| Blemishes | 92.8 96.5 92.3 | -0.6 | ±6.4 | Marginal | |
| Radiance | 93.0 89.5 83.4 | -9.5 | ±13.4 | Cannot support a verdict | |
| Redness | 75.8 71.7 88.7 | +12.9 | ±24.6 | Cannot support a verdict |