Square is the rarest face shape in this dataset, at 2.6% of 17,032 analyses. It is also the highest scoring, averaging 6.76 against an overall mean of 5.59. Those two facts together are why the shape question gets asked so often, and why the usual answer to it is incomplete: the advantage does not apply evenly to everyone with a square face, and the split is the actual finding.
How the four shapes distribute
The model returns a free-text description of face shape rather than picking from a fixed list, which produced 145 distinct labels across the dataset. A label is only counted as a shape when exactly one canonical shape word appears in it, so a description like "oval to oblong" is not forced into either bucket and lands in the mixed group instead.
| Shape | People | Share | Avg score | Avg symmetry | % scoring 7+ | % under 5 |
|---|---|---|---|---|---|---|
| Oblong | 7,566 | 44.4% | 5.55 | 5.59 | 8.7 | 30.1 |
| Oval | 7,199 | 42.3% | 5.66 | 5.68 | 5.4 | 14.3 |
| Round | 1,148 | 6.7% | 4.86 | 4.90 | 0.1 | 65.9 |
| Square | 445 | 2.6% | 6.76 | 6.74 | 47.9 | 15.1 |
A further 622 people, 3.7% of the dataset, received mixed labels and averaged 5.89. Another 52, 0.3%, fell outside these categories entirely.
So the most common face shape is oblong at 44.4%, with oval a close second at 42.3%. Between them they account for 86.7% of everyone. Round is 6.7%, and square is the least common at 2.6%.
The square result, and why it needs splitting by gender
The headline square figure is dramatic. Square faces average 6.76 and 47.9% of them score 7 or above, against 8.7% of oblong faces, 5.4% of oval faces and 0.1% of round faces. Nearly half of one group clears a threshold that only 8.1% of the whole dataset clears.
Split by gender and it stops being a story about shape.
| Shape | Male n | Male share | Male avg | Female n | Female share | Female avg |
|---|---|---|---|---|---|---|
| Oblong | 7,029 | 55.0% | 5.57 | 537 | 12.6% | 5.28 |
| Oval | 4,178 | 32.7% | 5.48 | 3,021 | 71.0% | 5.91 |
| Round | 699 | 5.5% | 4.74 | 449 | 10.5% | 5.04 |
| Square | 392 | 3.1% | 6.97 | 53 | 1.2% | 5.14 |
Male square faces average 6.97 against a male average of 5.55, which is a gap of nearly a point and a half and comfortably the largest shape effect in the data. Female square faces average 5.14 against a female average of 5.71, so the same shape sits slightly below the middle rather than well above it.
That interaction is the real finding here, and it is more interesting than the rarity fact. A square jaw is not a general-purpose advantage. It is scored as one within male faces and not within female ones.
The honest limitation
Only 53 women in 17,032 analyses received a square label. That is a small sample, and a small sample means the female square average of 5.14 carries real uncertainty - it could move meaningfully with a few hundred more cases. It is reported here because leaving it out would let the unsplit 6.76 stand as a claim about everybody, which the data does not support. But it should be read as a signal worth checking rather than a settled figure.
Two further caveats apply to the whole table. The gender split is uneven, 12,775 male against 4,257 female, so male patterns dominate any figure that is not split. And the shape labels come from a model reading a photograph, not from anthropometric measurement, so a borderline face can be described differently on different days.
What about the most sought-after shape?
This dataset cannot answer that. It measures what one model scores, not what people want or find desirable, and "sought after" is a preference question that would need entirely different evidence.
What it can say is that the scoring is not evenly spread across shapes, and that the round result deserves a note. Round faces average 4.86 with 65.9% scoring under 5 and only 0.1% reaching 7, which is the most concentrated result in the table. Some of that is genuine geometry and some is confounded: roundness in a photograph is affected by camera distance, lens focal length, angle and body composition, none of which the shape label separates out. A face photographed close up with a phone lens reads rounder than the same face photographed from further back.
That is worth knowing before treating a round label as a verdict. It is the shape most likely to be produced partly by the photograph.
Reading your own shape result
Face shape is a description, not a score. The averages above are group averages across thousands of people, and the spread inside every group is wide enough that your individual result is not predicted by your label. 15.1% of square faces score under 5, and 8.7% of oblong faces score 7 or above.
The practical use of knowing your shape is not the score attached to it. It is that shape genuinely informs grooming decisions, particularly hair, where the length and volume that suit a long face are different from what suits a wide one. Our guide to hairstyles for your face shape covers that side of it.
To find out which of the four you fall into, the face shape detector returns your label from a single photo, and the full shape breakdown including the mixed labels is in the most common face shape study.