A good score is not the same number on every feature, which is the thing almost every article on this subject gets wrong. Across 17,032 AI face analyses run between January and August 2026, the twelve measured features average between 5.97 and 5.30. Eyes sit highest at 5.97, skin lowest at 5.30, and nothing averages above 6.
So the short version: on any feature, above 6 is above average, a 7 is good, and an 8 is rare everywhere. But how rare a 7 is swings enormously by feature. 22.6% of people reach 7 on their eyes and only 7.7% reach it on their nose, which means the same number means very different things depending on where you scored it.
Every feature, with its real benchmark
This is the whole answer in one place, sorted from the highest-scoring feature to the lowest.
| Feature | Average | Above average is | % who score 7+ | Can you change it |
|---|---|---|---|---|
| Eyes | 5.97 | 6.0 or higher | 22.6% | Partly |
| Eyebrows | 5.94 | 6.0 or higher | 20.2% | Yes |
| Lips | 5.65 | 5.7 or higher | 16.4% | Partly |
| Symmetry | 5.62 | 5.7 or higher | 11.2% | Barely |
| Bone structure | 5.60 | 5.6 or higher | 16.7% | No |
| Jawline | 5.59 | 5.6 or higher | 17.4% | Partly |
| Facial harmony | 5.58 | 5.6 or higher | 12.5% | Barely |
| Cheeks | 5.42 | 5.5 or higher | 10.1% | Partly |
| Chin | 5.39 | 5.4 or higher | 10.6% | No |
| Forehead | 5.39 | 5.4 or higher | 7.9% | Partly |
| Nose | 5.36 | 5.4 or higher | 7.7% | No |
| Skin | 5.30 | 5.3 or higher | 16.6% | Yes |
Two things fall out of that table immediately. The averages are packed into a 0.67-point range, so no feature is scored generously and none is scored harshly. And the 7+ column is not packed at all: it runs from 22.6% down to 7.7%, nearly a threefold difference. That column, not the average, is what tells you whether a score is impressive.
The features almost everyone does well on
Eyes and eyebrows sit at the top and they sit there together. Eyes average 5.97 with 22.6% of people scoring 7 or above, and eyebrows average 5.94 with 20.2% doing the same. They are also the two features people are most likely to have as their personal best: eyebrows are someone's strongest feature 29.3% of the time and eyes 20.1% of the time, which between them accounts for roughly half of everyone.
The practical consequence is that a 6.5 on your eye rating is a good score but not a distinguishing one, because more than a fifth of people are above 7 there. If you are trying to work out which of your features is genuinely strong, the eyes and eyebrows results need discounting slightly against the rest.
Eyebrows have a second property that makes them the most useful feature in the whole table, and it gets its own section below.
The features where a low score is the normal outcome
The nose has the lowest 7+ rate of anything measured. Only 7.7% of people score 7 or above on their nose rating, and the average is 5.36. The forehead is barely better at 7.9%, and it is also the feature people are most likely to have as their weakest, at 15.0%.
This is worth saying plainly because the nose is the feature people are most anxious about. A 5.2 nose is not a flaw the model found. It is the median result, and 92.3% of people are also below 7 there. The same applies to the forehead rating, the chin rating at 10.6% and the cheekbones test at 10.1%.
Skin is the outlier in this group and needs separating out. It has the lowest average of all twelve at 5.30 and is someone's worst feature 38.0% of the time, which is by far the highest rate in the data. But 16.6% of people still score 7 or above on it, more than double the nose rate. Skin is not a feature where everyone is stuck low. It is a feature with an unusually wide spread, standard deviation 1.29, the widest measured, and people land all over it.
The two that actually move with effort
Only two of the twelve are marked as genuinely improvable, and they are skin and eyebrows.
Skin is the obvious one, and the numbers back it up. It has the widest spread of any feature, its 8+ scorers average 8.33 against a population mean of 5.30, and the people scoring under 5 overall average just 4.05 on skin, the lowest figure anywhere in the dataset. That is a 4.28-point gap between the top and bottom groups on a feature that responds to sleep, hydration, sun exposure, a basic routine and, not trivially, the lighting in the photograph. The skin analysis tool scores texture and clarity separately from everything else, which is the point: it is the one measurement where a different result next month is realistic.
Eyebrows are the underrated one. They average 5.94, they are the feature most likely to be someone's best at 29.3%, and grooming changes them within a week. If you are looking for the highest-return thing on this page, an eyebrow rating below 5.9 is the most fixable gap you are likely to have.
A few features are marked "Partly", and the honest reading of that is presentation rather than structure. The jawline rating responds to body composition and to camera angle. The lip rating and the cheeks respond to hydration and to lighting. What none of them respond to is effort applied directly to bone.
Which brings us to the three marked "No". Bone structure, chin and nose are fixed. The bone structure test is measuring the underlying architecture of your face, and it has the strongest 8+ average of anything in the data at 8.90 alongside a 4.58 average among low scorers. It is the biggest single differentiator and the one you have the least influence over. That is not a comfortable finding, but pretending otherwise would be worse.
The forehead is the feature that means the least
The forehead deserves a note of its own, because a low score there should worry you less than a low score anywhere else.
It has the smallest spread of any feature, a standard deviation of 0.78 against 1.29 for skin and 1.24 for jawline. It also has the lowest ceiling: people who score 8 or above overall still only average 7.66 on their forehead, the only feature in the whole table where top scorers do not clear 8. And the gap between the top group and the bottom group is 2.82 points, the narrowest measured.
Put those three facts together and the forehead is a compressed measurement. Almost everyone lands in a narrow band around 5.39, high scorers barely escape it, and the feature does comparatively little work in the final number. If your forehead is your weakest feature, and for 15.0% of people it is, you have found the least consequential weakness available.
How to read your own breakdown
Read each feature against its own column in the table, never against 10 and never against the other features. A 5.5 nose and a 5.5 eye score are not the same result: the nose is above the feature average and the eyes are half a point below it.
Then look at the spread rather than the peaks. Across everyone, the gap between a person's best and worst feature averages 1.38 points, 44.8% of people tie for their best and 58.0% tie for their worst. Most faces are fairly flat. If your own spread is much wider than 1.4 points, the low end is worth attention. If it is narrower, there is no single weak link to fix and the overall number is doing what it should.
The two features that summarise everything else are worth checking last. Symmetry and facial harmony correlate with the overall score at 0.95 and 0.99, so the face symmetry test and the facial harmony test are not really separate features. They are restatements of the whole result, which is why "Barely" is the honest answer to whether you can change them.
The full per-feature figures, including the correlations and the best-and-worst-feature rates, are in the which facial feature matters most study, and the overall distribution the feature scores sit inside is in the average face rating study.