A 4 out of 10 is below average, but not by as much as it feels. The mean face rating across 17,032 AI face analyses is 5.59 and the median is 5.3, so a 4 sits roughly a point and a half under the middle. It is also a common result rather than an unusual one: 4,230 people, or 24.8% of everyone analysed, scored somewhere in the 4.0 to 4.9 band.
More importantly, it is the band with the most room to move. The 4.0-4.9 group averages 1.75 points of estimated headroom - the largest of any band in the dataset - and an average potential score of 6.47, which is above what the average person scores today.
How many people score a 4 out of 10?
Almost exactly a quarter. The 4.0-4.9 band is the second-largest in the whole dataset, behind only 5.0-5.9. What is more useful is what sits underneath it: only 0.5% of people, 77 out of 17,032, scored below 4.0.
That matters because a 4 feels like the bottom of the scale and statistically is not. The distribution compresses hard at the low end - almost nobody is down there. A 4 is the lower quarter of a tight cluster, not an outlier, and the difference between a 4.2 and a 4.8 is a lot of people.
What does a 4 out of 10 actually look like, feature by feature?
This is where the data gets genuinely useful, because it separates what is fixable from what is not. Every analysis scores twelve features individually. Comparing what people scoring under 5 average on each feature against what people scoring 8 or above average shows exactly where the distance sits.
| Feature | Under 5 average | 8+ average | Gap |
|---|---|---|---|
| Jawline | 4.53 | 8.89 | 4.36 |
| Bone structure | 4.58 | 8.90 | 4.32 |
| Skin | 4.05 | 8.33 | 4.28 |
| Facial harmony | 4.55 | 8.66 | 4.11 |
| Chin | 4.43 | 8.48 | 4.05 |
| Symmetry | 4.74 | 8.54 | 3.81 |
| Cheeks | 4.50 | 8.25 | 3.75 |
| Eyes | 5.04 | 8.62 | 3.58 |
| Nose | 4.51 | 8.01 | 3.50 |
| Lips | 4.65 | 8.00 | 3.35 |
| Eyebrows | 5.11 | 8.27 | 3.16 |
| Forehead | 4.84 | 7.66 | 2.82 |
The top three gaps are within 0.08 of each other, so treat them as a near-tie rather than a ranking. The honest reading is that a low score is not caused by one feature - the distance is spread fairly evenly across all twelve.
But one line in that table is different from the rest. Skin is the lowest absolute score for this group at 4.05, the weakest of any feature, and it is the only one in the top five gaps that is fully improvable. Jawline and bone structure are largely fixed. Skin is not.
Can a 4 out of 10 improve?
More than any other band. Every analysis estimates a potential score alongside the current one, and the 4.0-4.9 band shows an average gap of 1.75 points, against 0.39 points for people already scoring 8.0-8.9.
The average potential for this band is 6.47. In other words, the model estimates that the typical person scoring a 4 has an achievable version of themselves that scores above the 5.59 population average - and 6.47 would land inside the top quarter, since only 25.1% of people score 6 or above.
That is not motivational framing, it is what the numbers say, and the mechanism is straightforward. A low score usually reflects several fixable inputs at once rather than one unfixable structural problem. Skin is the weakest feature for 38% of all people and the single most improvable thing measured. Eyebrows are the strongest feature for 29.3% and are also fully improvable.
Where the points are not
Being straight about the ceiling matters more here than anywhere else on the scale, because this is the band most exposed to people selling transformation.
Golden-ratio compliance for the 4.0-4.9 band averages 74.98%, against 89.03% for the 8.0-8.9 band. That gap is structural - it describes proportions, not presentation, and no routine changes it. The correlation between current and potential score across the whole dataset is 0.94, which is very high: potential tracks structure closely.
Concretely, of the 4,307 people scoring under 5, not one had an estimated potential of 8 or above. The highest potential in that entire group was 7.5. A 4 can realistically become a good 6. A 4 does not become a 9, and only 15 people out of 17,032 had a gap of even 3 points.
What a 4 does not mean
It does not mean unattractive to actual people. This is one model scoring facial geometry and presentation from a single photograph. It has no access to how you move, speak, dress or carry yourself, and those matter enormously to real attraction in ways no still image captures. If a 4 has you asking the blunter version of the question, the am I ugly test works through it on the same data - only 77 people out of 17,032 scored below 4.
It is also sensitive to the photo itself. Harsh overhead light, a tilted head, a wide-angle front camera held close, and a neutral-to-unhappy expression can each cost real points without anything about your face changing. Before concluding anything from a 4, it is worth retaking the photo at eye level in even, indirect light.
One further caveat applies to the whole dataset: people who use a face-rating tool are not a random sample of the population. These figures describe the people who showed up.
The complete distribution and percentile table is in the average face rating study, the headroom figures come from the potential vs actual study, and the feature-by-feature breakdown is from the which facial feature matters most study. If you want a realistic target rather than a guess, the face rating tool returns your current and potential scores together, and how much looksmaxxing can actually improve your looks covers what that gap means in practice.