S SoftMaxx ← Back to SoftMaxx
Methodology cornerstone

The 11 facial metrics that determine attractiveness (with research)

An anatomical breakdown of every category SoftMaxx scores, what the peer-reviewed literature suggests about each, and what is realistically changeable through softmaxxing.

SoftMaxx editorial · Updated May 2026 · ~13 min read

Facial attractiveness research consistently identifies a small, stable set of dimensions that predict perceived attractiveness across cultures. SoftMaxx measures 11 categories because that is the natural carve-up of the face into independent anatomical regions, and because each is informed by its own body of peer-reviewed anthropometric work (full bibliography on our science page). What follows is a category-by-category tour of what we measure, what the literature suggests is "in range," and what you can realistically change.

01Skin

What it measuresClarity, texture uniformity, tone evenness, hydration cues, periorbital condition.
Key sub-measurementsSkin homogeneity index, blemish density, redness distribution, undereye and oiliness signals.
How changeableVery high. Skincare, sleep, sun protection, and diet all move it within weeks.

Skin is the most visually dominant of the 11 categories. Surface qualities sit on top of every other feature you have, which is why a coarse skin read can drag down an otherwise strong face and a clean skin read can lift one. Work by Fink and colleagues on skin homogeneity as a health signal has repeatedly found that texture and tone evenness predict perceived attractiveness independently of facial shape, with isolated skin patches still carrying real signal even when separated from facial structure.

Even, hydrated skin with minimal redness tends to score in the upper range; visible acne, post-inflammatory hyperpigmentation, persistent oil sheen, or stark contrast between healthy and inflamed areas tends to score lower. Of all 11 categories, skin has the highest ratio of effort to result. Most users see meaningful movement within 60 to 90 days of a disciplined AM and PM routine plus daily SPF. For practical next steps, see our guide on how to look better.

02Eyes

What it measuresEye size, shape, canthal tilt, eye spacing, periorbital structure, hunter quotient.
Key sub-measurementsPalpebral fissure length, canthal tilt angle, intercanthal-to-alar ratio, vertical eye exposure, brow-to-eye distance.
How changeableMostly fixed bone structure, but soft-tissue framing (sleep, body fat, dermatology) can shift the read by a meaningful amount.

The eye region is anatomically dense. Within a small area you have orbital width, palpebral fissure length, the angle from medial to lateral canthus (the "canthal tilt"), and the relationship between the brow and the upper lid (the "hunter quotient"). Each has been studied in its own right in the orbital anthropometry and sexual-dimorphism literature, and each contributes independently to how an eye area reads.

Research suggests that a positive canthal tilt and a deeper-set, less exposed upper lid tend to read as more dimorphic in men. Eye spacing close to the canon (intercanthal width roughly equal to alar width) tends to read as harmonious; extremes in either direction move the score. Two related deep-dives: the canthal tilt guide and the hunter eyes guide. Soft-tissue improvements (better sleep, lower body fat, less periorbital oedema) can visibly tighten the eye area without touching the underlying bone.

03Brows

What it measuresDensity, shape, position relative to the orbital rim, distance from the lash line.
Key sub-measurementsBrow thickness, peak position, brow-to-eye distance, ridge prominence on profile.
How changeableHigh. Grooming, brow gel, lamination, and microblading all materially shift the read.

Brows sit at the top of the eye region and frame everything below them. The sexual-dimorphism literature is reasonably consistent that heavier, flatter, lower-set brows read as more masculine, and lighter, more arched, higher-set brows read as more feminine. For male users that usually means encouraging fullness across the body of the brow, keeping the line as flat as the natural growth allows, and resisting the urge to over-shape.

What changes in practice: shape, density, and tidy edges. Brow gel and a regular grooming routine can give a noticeably fuller-looking brow without altering the underlying hair count. For users with sparse or patchy growth, brow serums with peptide actives have a modest evidence base, and microblading is a practical longer-term option. What does not change without surgery is the underlying brow-ridge prominence (the bone shelf above the orbit), which is a separate read folded into the eyes and overall masculinity scores.

04Nose

What it measuresWidth-to-face ratio, projection from the face plane, tip rotation, dorsal profile.
Key sub-measurementsAlar base width, nasal length, nasolabial angle, dorsal hump presence, tip definition.
How changeableLow without surgery. Contouring helps women on camera; for men, framing changes (beard, brow density) have more leverage than the nose itself.

The nose is heavily studied in clinical anthropometry, mainly because rhinoplasty has produced a large normative literature. We measure alar width relative to intercanthal distance, nasal length relative to facial thirds, the nasolabial angle, and dorsal profile on the side-profile scan. Research-derived ranges suggest that an alar base close to the intercanthal width, a nasolabial angle of roughly 90 to 105 degrees for men, and a straight or very slight dorsal contour tend to read as harmonious in most reference populations.

The nose has the lowest soft-tissue leverage on the face. Cosmetic contouring lightly reshapes the read on camera, but in person makeup helps women far more than men. The honest framing is that the nose score is largely fixed without rhinoplasty, and effort is better spent on higher-leverage soft-tissue categories.

05Lips

What it measuresUpper-to-lower lip ratio, vermillion border definition, lip-to-face ratio, hydration.
Key sub-measurementsUpper lip thickness, lower lip thickness, lip seal, philtrum length, vermillion height.
How changeableModerate. Hydration, lip care, and dental work shift the read. Fillers shift it further.

Lips sit at the center of the lower face. The most-discussed ratio in the literature is upper-to-lower lip thickness, which tends to cluster around a 1:1.6 (upper:lower) range in faces rated as attractive on most reference panels, though there is real variance. The vermillion border (the sharp edge between lip and skin) reads more defined in younger, healthy lips and softer in older or sun-damaged ones; that single edge is one of the higher-leverage age signals on the face.

What you can change without filler: hydration, daily lip care, and removing chronic lip-licking and biting habits all materially improve the surface quality of the vermillion border. SPF on the lips is underrated. For users with significant volume loss or asymmetry that does not respond to soft-tissue work, fillers are an option with an established safety profile when done by a competent injector. Philtrum length is fixed without surgery.

06Jawline

What it measuresGonial angle, jaw width, mandibular projection, masseter definition, neck-jaw separation.
Key sub-measurementsGonial angle (target roughly 110 to 125 degrees in men), bigonial width, jaw line clarity, submental angle.
How changeablePartial. Bone is fixed. Soft-tissue framing (body fat, masseter development, posture) moves the visible read significantly.

The jawline is the structural anchor of the lower face. The angle where the body of the mandible meets the ramus (the "gonial angle") is the most-cited geometric measurement in male attractiveness work; research-derived ranges typically place a "harmonious" male gonial angle between roughly 110 and 125 degrees, with deeper angles reading as more dimorphic. We also measure jaw width relative to bizygomatic width, projection on the side profile, and the soft-tissue clarity of the jaw line from gonion to gnathion. For the deep-dive, see the gonial angle and jawline guide.

The bone is largely fixed past adolescence. The soft tissue sitting on top of it is not. Body composition (buccal and submental fat pads), masseter development, posture, and consistent sleep all noticeably tighten the jaw read without changing the skeleton. Most users who claim a "jaw transformation" from softmaxxing are reading a clearer outline of the same underlying bone, which is a real and durable change.

07Cheekbones

What it measuresZygomatic width, projection from the face plane, height relative to nasal mid-point.
Key sub-measurementsBizygomatic width, malar projection, cheekbone-to-jaw ratio, midface fullness.
How changeableLow without filler or implants. Body composition and contouring move the visible read modestly.

Cheekbones are skeletal. The zygomatic bone defines the widest point of the upper face, and its projection is one of the more discussed structural variables in both male and female attractiveness work. Research suggests that prominent, well-projected zygomas cluster with high attractiveness ratings in both sexes, though the effect interacts with overall facial fat distribution. Wide cheekbones on a lean midface read differently from wide cheekbones with a fully filled midface.

What you can change: midface body fat, which is most of what is visible on top of the zygomatic structure. Leaner body composition surfaces more underlying cheekbone shape, which is part of why "glow-up" transformations so often correlate with weight changes. On camera, contouring or judicious facial-hair shaping can lightly emphasize the malar plane. Without filler or implants the skeletal projection itself does not change, so users with a flatter native cheekbone should focus on the soft-tissue levers that actually move.

08Chin

What it measuresProjection, height, width, relationship to the lower lip.
Key sub-measurementsChin projection on side profile (Ricketts E-line context), chin height, mentolabial sulcus depth.
How changeableLow without genioplasty or filler. Submental fat reduction sharpens the read.

The chin is closely tied to perceived masculinity in men. Forward chin projection is one of the more consistent dimorphic signals in the attractiveness literature, and a recessed or "weak" chin is among the more common cosmetic concerns clinicians see in male patients. We measure chin projection relative to the Ricketts E-line on the side profile, chin height relative to the lower facial third, and the depth of the mentolabial sulcus.

What you can change without surgery: not the projection itself, but the read of the projection. Reducing submental fat sharpens the line between chin and neck and visually adds projection. Posture (chin neither tucked nor jutted) cleans up the photographic read. For users with a clinically meaningful chin recession that affects bite and breathing, genioplasty or orthognathic surgery is in scope but is a clinical decision, not a softmaxxing one.

09Ears

What it measuresProminence (angle off the head), size relative to face, vertical set.
Key sub-measurementsAuriculocephalic angle, ear length, helix position, lobe attachment.
How changeableLow without otoplasty. Hairstyle can frame or hide ears effectively.

Ears are the lowest-weighted of the 11 categories for most faces because they sit at the edge of the frame and read mostly through hair. The clinical norm for ear prominence is an auriculocephalic angle of roughly 20 to 30 degrees off the head, with anything significantly above that classified as prominent in the otoplasty literature. Ear length roughly tracks nasal length in clinical canon, though the perceptual relevance of small deviations is modest.

What you can change: hairstyle is the largest lever by a wide margin. Longer sides, ear-covering cuts, or strategically placed volume can fully neutralize the read of even fairly prominent ears. For users where the ears materially affect the frontal read, otoplasty is a relatively low-risk, well-established procedure with a high satisfaction rate in the published outcome literature.

10Hair

What it measuresDensity, frontal hairline integrity, contrast with skin, style fit.
Key sub-measurementsNorwood-equivalent stage, hairline shape (mature vs juvenile), part density, taper quality.
How changeableHigh. Topical and oral interventions, plus styling, all materially move the read.

Hair functions as a health signal and as a framing element for the entire face. Density and hairline integrity carry most of the weight, with style fit (cut quality, taper, length appropriate to face shape) adding meaningful additional signal. The evidence base for hair-density interventions in men is reasonably strong: finasteride and minoxidil both have decades of randomized-trial support for halting and partially reversing androgenic miniaturization, and surgical follicular unit extraction has a well-documented outcome literature.

What changes fast: style fit. A good cut from a barber who understands your hair type and face shape can move the hair read by half a category overnight, which is one of the higher-leverage single interventions across all 11. What changes slower but durably: density, via consistent topical or systemic intervention if clinically appropriate. What does not change is the underlying follicular pattern past a certain stage, which is why earlier intervention has materially better outcomes.

11Symmetry

What it measuresBilateral balance of all features across the vertical midline.
Key sub-measurementsEye height differential, oral commissure level, alar base symmetry, ear height differential, midline deviation.
How changeableMostly fixed. Posture, dental work, and asymmetric muscle development account for most movable asymmetry.

Symmetry is the most-studied single variable in the facial attractiveness literature, with Rhodes and colleagues' meta-analytic work being the most-cited synthesis. The headline finding is consistent: bilateral facial symmetry positively predicts perceived attractiveness, with the standard interpretation that symmetry signals developmental stability. The effect is real and replicable across populations, though modest in magnitude. Small everyday asymmetries (within roughly two to three millimeters at the relevant landmark pairs) are essentially imperceptible to viewers.

What you can change: not much of the skeletal asymmetry, but a meaningful fraction of the soft-tissue and habitual asymmetry. Chronic side-sleeping, dominant chewing side, persistent neck rotation from screen setup, and uncorrected dental midline deviation all contribute and respond to intervention. Severe asymmetries usually have a clinical cause and should be evaluated by a professional.

12How SoftMaxx scores all 11

Putting the 11 together is where most face-rating tools fail. A simple average across categories produces a misleading overall number, because feature weights are not uniform and a strong feature cannot fully compensate for a clearly weak one. SoftMaxx computes more than 50 ratio measurements from a 478-landmark facial geometry pass on your photo, combines those with an AI vision read on each of the 11 categories, and produces sub-scores per category. The final overall score is a weighted geometric mean of the sub-scores, calibrated against celebrity benchmark photos so the average man scores about 5 out of 10.

The geometric-mean choice penalizes clearly weak categories appropriately. A face that is a 7 in nine categories and a 2 in one should not score a 6.5 overall; it should score lower, because the visible deficit pulls the read down in a way arithmetic averaging does not capture. For the full technical breakdown of the scoring pipeline, see our methodology page, and to see your own 11 sub-scores, you can run a scan in the SoftMaxx app in about thirty seconds.

Want to see your own 11 scores? The free tier of SoftMaxx covers the front scan, your overall score, all 11 category sub-scores, and a basic protocol. Pro at $19.99/mo unlocks the full protocol, the AI coach, rescans, and the side-profile measurements. Elite at $39.99/mo adds persistent memory and tracking; Elite Annual is $150/yr.

Run your scan at softmaxx.io/app.

Common questions

Why 11 categories and not 5 or 20?

Facial attractiveness research tends to cluster features into roughly 11 independent anatomical regions. Fewer categories lose fidelity (collapsing skin and texture into "overall" hides a quarter of the signal); more categories create redundancy (separating lower-lip thickness from upper-lip thickness rarely tells you anything that "lip proportion" does not already cover). Eleven is the natural carve-up of the face into regions that move independently.

Which of the 11 categories matters most?

Skin and eyes tend to carry the heaviest weight in cross-cultural attractiveness research because they function as health signals. Skin homogeneity in particular has shown effects even when isolated from facial shape. After that, jawline, symmetry, and hair cluster as the next tier. The exact ranking depends on viewer, context, and the sex of the face being rated.

Can I improve every category?

No. Skin, hair, brows, lips, and the soft-tissue framing of the jawline (via body composition and posture) are highly changeable through consistent softmaxxing. Nose shape, cheekbone projection, chin projection, and ear position are skeletal, so only surgery materially changes them. Symmetry is mostly fixed. The honest split is roughly half changeable, half not, with the changeable half driving most realistic improvement. For a primer on what softmaxxing covers, see what is softmaxxing.

How does SoftMaxx know what "ideal" looks like?

Scoring is anchored to two reference sets: peer-reviewed anthropometric ratio research from the orthodontic, maxillofacial, and facial-attractiveness literature, and a benchmark set of celebrity reference photos with consensus rater ratings. Together those define what each ratio range tends to score, and the population baseline is calibrated so the average man scores about 5 out of 10. The full breakdown of which research streams we draw on lives on our science page, and if you have specific questions about scoring, our FAQ covers the common ones.

Is the same "ideal" true across cultures?

Largely yes. Cross-cultural attractiveness studies consistently find high agreement on which faces are rated as attractive, with most variance concentrated in specific features (such as skin tone preference or eye shape) rather than the underlying structural ratios. SoftMaxx uses population-aware norms where the literature supports them and flags uncertainty where it does not.

Continue reading

If you want to go deeper on the specific structural measurements behind a few of the highest-leverage categories, these three guides cover the relevant anatomy and the practical question of what is and is not changeable.