Reference
The 1-10 attractiveness scale, with actual numbers under it
Everyone uses the 1-10 scale and nobody agrees what a 7 is. Here is what the bands mean on a measured geometric scale, where real faces land, and why the useful range is much narrower than the convention suggests.
The bands used on this site
Each band is a range on the 0-100 composite, shown out of 10. The breakpoints are the percentiles of the distribution the scoring engine was calibrated against - 29,927 measured faces - so the population claims are literally true rather than decorative.
| Band | Score | Population | What it describes |
|---|---|---|---|
| Elite | 8.7+ | Ahead of 95% of scans | Top-of-distribution balance across every measured dimension. |
| Striking | 8.3 - 8.7 | Ahead of 85% of scans | Strong overall harmony with one or two standout features. |
| Refined | 7.8 - 8.3 | Ahead of 65% of scans | Above-average balance, with clear measurable room above it. |
| Balanced | 6.0 - 7.8 | No claim made | A solid, even baseline - no single dimension drags it down. |
| Distinctive | 0.0 - 6.0 | No claim made | Individual features the geometry scores low and people don't. |
Why almost everyone lands between 6 and 8
The informal 1-10 scale imagines attractiveness spread evenly across ten steps, with plenty of people at 3 and plenty at 9. Measured geometry does not look like that at all. Human faces are strikingly similar on symmetry, thirds and fifths - the whole species converges on a narrow band, and the differences that feel enormous in a mirror are usually a few percent of face width.
That is why the middle of this scale is crowded and the ends are nearly empty. A face measuring 6.4 and one measuring 8.1 are not two different species of face; they differ by an amount that a change of camera angle can partly erase.
What the scale cannot carry
A geometric scale measures shape. The informal scale people use in conversation is trying to summarise a social response, which includes expression, grooming, style, confidence, voice, height, context and familiarity - none of which appear in a landmark mesh.
These two scales are frequently treated as the same thing, and they are not. A number from this site tells you where your facial geometry sits among scanned faces. It does not tell you what number a person would assign you, and no measurement can.
Questions people ask
- What is the 1-10 attractiveness scale?
- It is an informal rating convention, not a measurement standard. It has no agreed definition, no reference points and no consistency between two people using it, which is exactly why the same face gets called a 5 by one person and an 8 by another. On this site the 0-10 figure is a rescaled geometric composite with published breakpoints, which makes it consistent - but it is still not the same quantity the informal scale is gesturing at.
- Where do most people score?
- On measured geometry, the great majority of faces land between 6.0 and 8.0 out of 10. The distribution is narrow because human faces genuinely are similar on these measures: symmetry, thirds and fifths vary far less between people than the 1-10 convention implies.
- What counts as a good score?
- 8.3 and above sits in the top 15% of scans on this site, and 8.7 and above in the top 5%. Below 6.0 the bands deliberately carry no percentile claim, because telling someone where they rank at the bottom of a distribution is not information anyone benefits from.
- Why does the scale not run the full 1 to 10?
- Because faces do not. Scoring 3 out of 10 on this composite requires geometry so far outside the measured distribution that it is almost always a bad photograph rather than a face - a heavy head turn, a partial crop, or a face at a steep angle. A scale whose ends are unreachable in practice is being honest about its resolution.
- Is the male scale different from the female scale?
- No. The same breakpoints apply to every face. Symmetry and proportion measures are not sex-specific, and splitting the scale by sex would mean inventing separate ideals that the underlying measurements do not support.
See where your own geometry lands
The test reports your band, the composite, and all six component scores it was built from.