Across 17,032 AI face analyses, the average gap between a person's current score and their estimated potential is 1.31 points out of 10. Most people — 78% — have between 1.0 and 2.0 points of realistic headroom. Only 4.4% have 2 points or more, and out of all 17,032 people, not a single one had a gap of 4 points.
That is the honest answer to a question the looksmaxxing world usually answers with a vibe. Search it and you will be told "1-3 points" by a grooming brand, "it is over" by a forum, and "anything is possible" by someone selling a course. None of them show their working. We can, because every analysis on FaceRating.ai estimates a potential score alongside the current one, and we have 17,032 of them on a single scoring model. Those forum arguments are usually held on the PSL scale rather than out of 10, and the PSL rating test maps a measured score onto those bands if that is the vocabulary you are used to.
How many points can looksmaxxing realistically add?
The distribution is much tighter than the discourse suggests. Here is the full spread of current-to-potential gaps across all 17,032 analyses.
| Headroom | Share of people |
|---|---|
| Under 0.5 points | 2.8% |
| 0.5 – 1.0 points | 14.4% |
| 1.0 – 1.5 points | 36.7% |
| 1.5 – 2.0 points | 41.7% |
| 2.0 points or more | 4.4% |
The median gap is 1.3 points and the standard deviation is just 0.42, which is the statistically interesting part: the amount of improvement available is remarkably consistent from person to person. Whatever your starting point, the model rarely thinks you have less than a point available, and it almost never thinks you have more than two.
It is worth being precise about what "potential" means here. It is not a prediction that you will reach that score, and it is not a promise. It is the model's estimate of where your face lands if the changeable inputs — skin, grooming, hair, body fat, the photograph itself — are in good shape, with bone structure held fixed. It is a ceiling, not a forecast.
Who has the most room to improve?
Counterintuitively, the people with the most headroom are the ones scoring lowest. This is the opposite of how looksmaxxing culture usually frames it, where low scorers are told they are stuck and high scorers are sold marginal gains.
| Current score | Average headroom | Average potential |
|---|---|---|
| Below 4.0 | 2.02 | 5.17 |
| 4.0 – 4.9 | 1.75 | 6.47 |
| 5.0 – 5.9 | 1.36 | 6.66 |
| 6.0 – 6.9 | 0.87 | 7.53 |
| 7.0 – 7.9 | 0.73 | 8.33 |
| 8.0 – 8.9 | 0.39 | 8.84 |
| 9.0 and above | 0.32 | 9.51 |
Someone scoring in the 4.0-4.9 band has, on average, 1.75 points available. Someone already at 8.0-8.9 has 0.39. That makes basic sense once you think about what the potential score is measuring: if you are already scoring 8.5, your skin is probably clear, your hair suits you and you know how to take a photograph. There is very little left on the table. If you are at 4.5, several of those things are usually not true yet, and each one is fixable.
The practical reading: if you are below average, the effort is worth more to you than it is to anyone else. That is genuinely encouraging, and it is the opposite of what the blackpill corner of the internet tells people.
Can you go from a 5 to a 9?
No. This is the clearest finding in the dataset and it deserves to be stated plainly.
Of 17,032 people, 15 had a gap of 3 points or more. That is 0.09%. Nobody at all had a gap of 4. And of the 4,307 people who scored below 5, not one had an estimated potential of 8 or above — the highest potential in that entire group was 7.5.
The correlation between current and potential score is 0.94, which is very high. Potential tracks structure. Grooming, skincare and photography move you within a band; they do not move you between bands. A 5 can become a good 6, sometimes a 7. A 5 does not become a 9, and no amount of routine changes that.
That sounds harsh written down, but in practice it is the more useful message. A realistic target you can actually hit beats an impossible one you will quit on, and 1.31 points is a visible difference in every photo you appear in for the rest of your life.
What should you actually work on first?
The same dataset answers this, and the answer is unusually consistent. Skin is the weakest feature for 38% of people — two and a half times more common than the next-most-frequent weak point — and it is the lowest-scoring of the twelve features on average at 5.30 out of 10.
It is also the single most improvable thing on the list. Bone structure is fixed. Skin is not. If you want the highest return per unit of effort, and you have not already sorted out sleep, sun protection and a basic routine, that is where the points are.
The second thing worth knowing is that faces are more even than people assume. The average gap between a person's best and worst feature is only 1.38 points, so most people do not have one catastrophic flaw dragging everything down — they have twelve features clustered close together. That matters for strategy: broad maintenance beats obsessing over a single feature, which is the failure mode looksmaxxing forums encourage.
Why the ceiling exists at all
A potential score has to be bounded by something, or it is just flattery. In this model it is bounded by the parts of a face that do not respond to effort — proportions, bone structure, the underlying geometry. Those are measured, held constant, and everything else is allowed to reach its realistic best.
This is also why the numbers here are lower than the ones you will see quoted elsewhere. A tool that tells everyone they can gain three points is not measuring anything; it is selling hope. The measured answer is smaller, duller and considerably more useful.
The honest limitations
These are one model's scores, not an objective verdict on anybody. Attractiveness is subjective and cultural, and no score out of 10 changes how a specific person sees you.
The potential estimate is also a model output rather than a measured outcome — we have not followed 17,032 people for two years to see who actually closed their gap. What we can say is what the model consistently estimates as achievable, across a large sample, on one scale. That is a genuinely different thing from a forum estimate, and it is the closest thing to real data this question currently has.
One more caveat worth stating: people who use a face-rating tool are not a random sample of humanity. This measures the people who showed up.
If you want to see where you actually sit, the face rating tool gives you both numbers — current and potential — from one photo, free once a day, and the looksmax AI analysis presents the same pair with the features that respond to effort separated from the ones that do not. The full dataset behind this article, including the distribution tables and methodology, is in the potential vs actual study.