A 7 out of 10 is a high score. Only 8.1% of 17,032 AI face analyses reached 7.0 or above - roughly 1 in 12 people - against a measured average of 5.59. If you scored a 7, you are well inside the top tenth of everyone who has taken the test.
It is also the point on the scale where the reason for the score changes character, which is the more interesting finding.
How rare is a 7 out of 10?
824 people scored in the 7.0-7.9 band specifically, which is 4.8% of the dataset. Stack the bands and the shape is clear: 8.1% score 7 or above, 3.3% score 8 or above, and 0.2% reach 9.
The percentiles put a finer point on it. The 90th percentile of all scores is 6.8, the 95th is 7.8 and the 99th is 8.8. So a 7.0 is already past roughly 92 people in 100, and a 7.8 is at the top-5% line.
Why do 7s feel more common than they are?
Because most rating you encounter is social, and social rating inflates. Friends, dating apps and casual "rate me" threads cluster around 7 and 8 because nobody wants to hand a real person a 4. A calibrated model has no such incentive, so its 7s are far scarcer than conversational 7s.
This is also why cross-tool comparisons mislead. Some rating tools report average scores above 7, which tells you about their scale rather than about faces. A 7 here and a 7 there are not the same object, and neither is comparable to a 7 from a friend. If the question underneath the number is really whether you are pretty rather than what the digit means, the am I pretty test answers it against the same 17,032 analyses.
What it actually takes to score a 7
An overall 7 is not one excellent feature carrying the rest. Every analysis scores twelve features separately, and the share of people who clear 7 on each one individually shows why the overall figure is so much lower.
| Feature | Share scoring 7 or above |
|---|---|
| Eyes | 22.6% |
| Eyebrows | 20.2% |
| Jawline | 17.4% |
| Bone structure | 16.7% |
| Skin | 16.6% |
| Lips | 16.4% |
| Facial harmony | 12.5% |
| Symmetry | 11.2% |
| Chin | 10.6% |
| Cheeks | 10.1% |
| Forehead | 7.9% |
| Nose | 7.7% |
Not one feature clears 7 for more than 22.6% of people. Eyes are the most generous, the nose the least - fewer than 1 person in 12 scores 7 or above on it.
An overall 7 requires most of these at once, in the same face. That is why 8.1% clear 7 overall even though 22.6% clear it on eyes alone. It is a conjunction, not a maximum, and conjunctions are rare.
The structural step at 7
Something measurable changes at this boundary. Golden-ratio compliance - how closely facial proportions track the classical ratios - averages 78.07% for the 6.0-6.9 band and 82.28% for the 7.0-7.9 band, a jump of 4.21 points.
For scale, moving from the 4 band all the way to the 6 band raises compliance by 3.09 points in total. The single step from 6 to 7 is larger than the entire climb beneath it.
The practical reading: below a 7, score differences are substantially about presentation - skin, grooming, hair, body composition, photography. From 7 upward, structure is doing much more of the work, and structure does not respond to routine.
Face shape and the 7 threshold
One shape clears this bar far more often than the others. Among square faces, 47.9% score 7 or above, against 8.7% of oblong faces, 5.4% of oval faces and 0.1% of round faces.
Two caveats keep that honest. Square is only 2.6% of the dataset - 445 people - so it is a small group. And the effect is essentially men-only: male square faces average 6.97 against a male average of 5.55, while the 53 female square faces average 5.14, which is below the female average of 5.71.
What is left above a 7?
Less than lower down, which is the pattern throughout the data. The 7.0-7.9 band averages 0.73 points of headroom with an average potential of 8.33 - so the realistic move is into the top 3%, not beyond it.
At this level the fundamentals are usually already handled, and the remaining points come from consistency rather than change: keeping skin in good condition, maintaining body composition, and not undermining a strong face with a bad photograph. The gap between someone's best and worst photo is frequently larger than 0.73 points, which means at this level the camera is often the biggest single variable left.
What will not happen is a 7 becoming a 10. Golden-ratio compliance tops out at 96% across all 17,032 faces, and nobody in the dataset showed a 4-point gap between current and potential.
The full percentile table is in the average face rating study, the headroom figures come from the potential vs actual study, the per-feature data is from the which facial feature matters most study, and the shape figures are from the face shape study. See your own two numbers with the face rating tool, get your percentile from the attractiveness test, or read is an 8 out of 10 a good face rating for what the band above looks like.