Face Rating Apps Give You a Number. Here's What They're Not Measuring.

A face rater takes a photo and returns one output: a score out of 10, a percentile, sometimes a letter grade. That number looks precise because it resembles data, but a single score is an average, and averages hide their inputs. Underneath almost every face rating sit several genuinely separate, independently measurable facts, symmetry, proportion, and jaw structure among them, each with its own research literature and its own citable threshold. None of those show up individually in a rater's output. All three can be measured and sourced on their own, and seeing them separately tells you something a single number cannot.

Definition

Three measurements a single score collapses

1. Symmetry

Symmetry is one of the few factors with consistent support in the controlled attractiveness literature. Little, Jones, and DeBruine's 2011 review in Philosophical Transactions of the Royal Society B, a standard reference in the evolutionary psychology literature on this topic, summarizes decades of controlled work and identifies symmetry as one of three consistent predictors of attractiveness judgments, alongside averageness and sexually dimorphic features. A face rater's single score cannot tell you whether symmetry is pulling the number up or down, because symmetry was never reported on its own. It was averaged in with everything else before you saw the output.

2. Facial proportion

The classical rule taught for facial proportion says the face divides into three visually equal sections, upper, middle, and lower. Farkas, Hreczko, Kolar, and Munro tested that rule directly in a 1985 study in Plastic and Reconstructive Surgery, measuring 153 young adult subjects, and found the lower third ran longer than the middle third on average, by 9mm or more in 68 percent of subjects. Proportion is not a single fact either. It has an upper-to-middle relationship and a middle-to-lower relationship, and a rater's one number cannot show which of those is close to typical and which is not.

3. Jaw structure

"Strong jawline" reduces, on the skeletal side, to one measurable value: the gonial angle, the angle where the ramus of the mandible meets its body. Riolo's 1974 cephalometric norms put the population mean around 124 degrees for adult males and 122 for adult females. Mommaerts' 2016 survey in the Journal of Cranio-Maxillofacial Surgery found the rater-preferred angle sits close to that population mean rather than far below it, meaning the widely repeated idea that attractiveness requires an extreme angle is not what the data shows. A face rating score has no way to surface a specific angle in degrees, or to say whether it sits inside or outside a published range.

Why this isn't just semantics

A rating is a judgment: it places a face on a scale someone else defined, with no visible method behind the placement. A measurement is a fact about a specific structure: an angle in degrees, a distance in millimeters, a ratio against a population reference, each traceable to a defined method and a citable source. The difference is not tone, it is falsifiability. A rating can only be agreed with or disliked. A measurement can be checked against the paper it came from, or against a caliper. Reframed that way, the goal was never a better rating. It was measurement in the first place, with rating as the shortcut that skipped it.

How to tell what you're actually looking at

• Check the output unit. A score out of 10 or a percentile with no stated basis is a rating. A distance in millimeters, an angle in degrees, or a ratio against a named reference range is a measurement.
• Ask for the source. A measurement tool can point to the specific paper behind each threshold, the way the studies above can be looked up directly. A rater that cannot name a source is not withholding a trade secret, it does not have one to show.
• Count the outputs. One number means everything was averaged before you saw it. A breakdown across separate categories, symmetry, proportion, projection, and the rest scored individually, means you can see which one is actually driving the result.
• Upload the same photo twice. A deterministic, landmark-based measurement returns the same numbers. A single-score rating model frequently does not.

What this looks like in practice

Facet does not produce a single face rating. A scan runs 468 facial landmark points through a deterministic scoring engine, then reports symmetry and proportion inside the overall harmony module and jaw structure inside the jawline module, each checked against a threshold drawn from a specific peer-reviewed source, the same standard cited above. The result is a breakdown across 10 modules with a citation attached to every number, not one score standing in for all of them. That structure, and how it differs from the closest subscription-based clinical alternative on the market, is covered at [Facet's homepage](/) and in the direct comparison at [Facet vs QOVES](/alternatives/qoves).

The fastest way to see what a face rater's single number is actually averaging together is to look at the measurements underneath it. A complete example scan, every module and every cited threshold filled in, is posted at [/sample-report](/sample-report). Run your own photo through it to see symmetry, proportion, and jaw structure reported separately instead of collapsed into one score.