Here is the honest answer up front. Across 23,714 real analyses on FaceRating.ai, the average face rating is 5.96 out of 10 — but the average potential score, the model's estimate of where the same face could realistically land by improving only the controllable things (skin, grooming, hair, styling), is 7.24. That is a gap of about 1.28 points, and it is not spread evenly across the face. In our data, the lowest-scoring features are the nose (5.74 average), skin (5.77), forehead and chin (5.80 each), and cheeks (5.83) — and of those, skin is by far the most improvable without surgery, while the nose, chin, and forehead are mostly structural. This guide walks through the feature data, separates what you can genuinely change from what you largely cannot, and sticks to interventions with actual evidence behind them. No "bone smashing," no miracle routines, no guarantees.
Where the 1.28-point gap comes from
Alongside every current score, our model estimates a potential score: where the same face could realistically land by improving the things a person can control — skin, grooming, hair, styling — with no surgery assumed. Across all analyses in our average face rating score study, the average potential is 7.24 against an average current score of 5.96. Two honest framings matter here. First, the potential score is a model estimate, not a promise: nobody can guarantee your score will move by any specific amount, and we won't pretend otherwise. Second, the size of the gap is itself useful information — about a point and a quarter is what the model considers realistic for a typical face. If a product promises to take you from a 5 to a 9, the data says it is selling you something other than the truth.
The feature data: what actually scores lowest
Averaged across every analysis, the twelve measured features range from 6.32 (eyes) down to 5.74 (nose). The table below is the practical heart of this article, because the low scorers split cleanly into features you can work on and features you mostly cannot.
| Feature | Avg score | Improvable? | What actually helps |
|---|---|---|---|
| Eyes | 6.32 | Partly | Sleep and reduced puffiness; brow and lash grooming change how the eye area reads |
| Eyebrows | 6.25 | Yes | Shaping and grooming — the cheapest, fastest cosmetic lever |
| Lips | 6.04 | Partly | Hydration and basic lip care; the underlying shape is structural |
| Symmetry | 6.03 | Barely | Largely set during development; styling can shift perceived balance |
| Bone structure | 6.02 | No | Fixed without surgery |
| Jaw | 5.99 | Partly | Lower body fat reveals existing definition; posture and puffiness affect how it photographs |
| Facial harmony | 5.98 | Barely | Downstream of overall proportions |
| Cheeks | 5.83 | Partly | Body-fat changes alter facial adiposity |
| Chin | 5.80 | No | Structural; beard styling can change the outline, not the bone |
| Forehead | 5.80 | Partly | A fringe or hairstyle changes its apparent proportion, not its size |
| Skin | 5.77 | Yes — most improvable | Daily SPF, retinoids, consistent skincare, sleep |
| Nose | 5.74 | No | Structural; only photo angle and lighting change how it reads |
Skin: the lowest improvable score on the board
Skin averages 5.77 — the second-lowest feature score in the entire dataset, and the lowest one that is genuinely changeable. The evidence that skin condition affects how faces are judged is solid: in a 2006 study, Fink, Grammer, and Matts showed that skin colour distribution alone — with face shape and surface topography held constant — changed how old, healthy, and attractive the same female faces were rated (later work extended similar findings to male skin). The practical levers are unglamorous and well-established: daily broad-spectrum sunscreen to prevent further photodamage, topical retinoids (one of the most-studied ingredient classes in dermatology for photoaged skin — introduce them gradually, ideally with medical guidance), basic moisturisation, and consistency over months rather than weeks. What we will not tell you is that any routine guarantees a specific score change. Skin responds at different rates for different people, and conditions like persistent acne deserve a dermatologist, not a face-scoring app.
Jawline and cheeks: mostly a body-composition story
Jaw (5.99) and cheeks (5.83) sit in the lower-middle of the table, and for many people the biggest non-surgical influence on both is body fat. Facial adiposity — how much fat the face carries — measurably changes how faces are perceived: Coetzee, Perrett, and Stephen (2009) found that perceived facial weight predicted both attractiveness and health judgments. So if you carry excess body fat, general fat loss tends to show in the cheeks and jawline. The honest caveats: if you are already lean, this lever is already spent, and the underlying jaw and chin shape is bone — no exercise changes a skeleton. Chewing-resistance training can build the masseter muscle, which is a soft-tissue change to how wide the jaw area looks, not a structural one. Beyond that, the things that affect how this region photographs are mundane: posture, sleep, and the puffiness that follows salt and alcohol.
Hair and brows: changing perceived proportion
Eyebrows already average 6.25 — the second-highest feature — but they stay on the improvable list because small grooming changes shift how the whole eye area and facial balance read. Hair is the bigger tool: a cut cannot change your measured bone structure, but it redraws the visible outline of the face and the apparent balance of the facial thirds. The forehead score (5.80) is a good example of a feature whose apparent proportion — though not its actual size — responds to a fringe or a different parting. Our guide to the best hairstyles for your face shape covers the balancing logic cut by cut, and it is the highest-leverage place to start if your proportions run long or wide.
What is structural — and the honest limits
The bottom of the feature table is dominated by structure: the nose (5.74, the lowest average of all twelve features), the chin (5.80), and bone structure (6.02) are fixed short of surgery, and symmetry (6.03) is largely set during development. This is worth stating plainly, because the looksmaxxing internet is full of claims to the contrary — "mewing" your way to a new jawline, "bone smashing," facial exercises that reshape the skull. None of it has credible evidence. It is also worth keeping the stakes in proportion: the attractiveness literature (Rhodes, 2006) finds that even the biggest structural factors — symmetry, averageness — have moderate effects on how faces are rated, not decisive ones. That cuts both ways: your structural "flaws" matter less than the forums claim, and so would surgically changing them. If you are seriously considering surgery, that is a conversation for a board-certified surgeon about your anatomy and health — not something to decide from an app score, ours included.
Photo quality: not your face, but your measurement
One more gap-closer costs nothing: the photo itself. Uneven lighting creates false shadows that read as asymmetry, a tilted head compresses one side of the face, and filters or heavy makeup move apparent landmark positions. Fixing these does not improve your face — it improves the measurement of your face, which is often where a surprising chunk of a low score lives. Use even light, a straight-on angle, a neutral expression, and no filter, and run the analysis under the same conditions each time so you are comparing like with like.
A realistic order of operations
If you want to close your own gap, this order front-loads the highest-evidence, lowest-cost changes. One: take a clean baseline photo and note your per-feature scores, not just the overall number. Two: start daily SPF and a basic skincare routine — skin is the lowest-scoring improvable feature on average. Three: sort out brows and general grooming. Four: choose a hairstyle that balances your specific proportions rather than someone else's. Five: if you carry extra body fat, general fat loss will show in the cheeks and jaw. Six: re-test after eight to twelve weeks under identical photo conditions. Expect movement measured in fractions of a point. The average gap between current and potential scores is 1.28 points — that is the realistic ceiling for most people, and closing even part of it is a genuine, visible change.
The takeaway
The data says most people have real but bounded headroom: on average about 1.3 points between where they score and where the model estimates they could land using only controllable factors. The route runs through skin, grooming, hair, body composition, and photo quality — not through chasing structural change. Read your own feature breakdown, work the improvable end of the table, and interpret the result against real numbers — our guide to what counts as a good face rating score shows the full distribution. And keep the frame honest in both directions: a face rating measures geometry on the day of one photo. Improving it is a reasonable project; it was never a measure of your worth.