Facial harmony: what your overall score really measures

Every Facet report opens with a single number: the overall harmony score, on a 0.0 to 10.0 scale. It is the most-screenshotted number in the product. It is also the most misunderstood.

The overall score is not an attractiveness rating. It is an aggregation of how closely your face fits the published clinical baselines across 10 modules, weighted by which modules influence first-impression read of a face.

The formula in plain English

The current overall score uses a weighted formula called "holistic-first-impression-v1." Module weights reflect how strongly each module influences the standard first-second read of a face. Some modules carry more weight than others:

• Skin: high weight. Skin condition is the first feature most viewers consciously or unconsciously assess.
• Eyes (canthal tilt) and Eyebrows: high weight. Upper-third features anchor expression and read instantly.
• Jawline and Nose: medium weight. Visible at all angles, but less dominant than upper-third.
• Lips and Cheeks: medium weight.
• Hair (or Beard for males): contextual weight. Strong when present, less load-bearing on overall harmony.
• Overall harmony (the meta-module): aggregation of the rest, not weighted in itself.

Each module score is itself a function of clinical parameters compared against ethnicity-keyed gender-typical thresholds. The final score is a weighted average mapped to a 0 to 10 band.

What the score does not measure

• Attractiveness. There is no published study that maps a single facial-parameter aggregate cleanly to a perceived attractiveness rating in a way that would justify calling Facet's score an attractiveness measure.
• Photogenicity. The score depends on a clinical-quality scan and is robust to small expression and lighting variance. It does not predict how a face will read in a smiling iPhone selfie under tungsten light.
• Confidence, charisma, expressivity, personality. None of these are extractable from a still photograph.

How to read your score

A useful framing: read the score against your own baseline, not against other people. A user with a 7.4 in May, a skincare protocol in place, and a 7.7 in November has done something measurable. A user comparing 7.4 to a friend's 8.1 is comparing two different faces against thresholds that do not directly compete.

The score's utility is that it changes when you change. Lower skin homogeneity from sun damage will drop the score over time. A consistent retinoid protocol will lift it. The number tracks the underlying state of the face.

Where the score breaks down

The score has known limits. Atypical faces (post-procedure, partial facial paralysis, asymmetric scarring) interact unpredictably with the threshold math. Mixed-ethnicity users using the averaged Mixed / Other baseline will get a less precise score than users matched to a specific population baseline. The current weighting scheme was set conservatively and will be revised as more user data accumulates.

If your score reads as off, the right move is to look at the per-module scores, see which one is dragging the average, and decide if that module is the one you actually care about. The overall is a summary. The breakdown is the report.