What an AI headshot audit actually measures
A headshot is read by a recruiter in well under a second. The AI ranker on this page is not guessing what looks “nice”. It scores the same structural inputs that decades of perception research have repeatedly tied to first-impression judgements of competence, trust, and warmth. Those inputs sit in five buckets. First, facial geometry: symmetry, proportions, jawline definition. Second, eye and gaze: pupil visibility, squint level, direction of look. Third, expression: smile genuineness, lip-corner asymmetry, brow tension. Fourth, framing: head size in the frame, crop tightness, vertical centring. Fifth, lighting and skin: contrast balance, shadow direction, perceived clarity.
For a LinkedIn or professional context the weighting is tuned toward traits recruiters consciously screen for: a clear primary subject, relaxed eye contact, no visual noise behind the head, and a smile with engaged eyes rather than a flat closed-mouth pose. Each photo gets a per-axis score plus a single composite, so you can see why one shot wins. The source literature behind those metrics is catalogued on the research bibliography page.
How RealSmile scores headshots versus general selfies
The same face can look great in one context and wrong in another. A high-energy beach selfie that performs well as a Hinge primary is often a poor LinkedIn lead, and a buttoned-up corporate headshot rarely opens dating conversations. RealSmile addresses this with two different scoring profiles riding on a shared geometric backbone.
In headshot mode the model up-weights perceived competence and trust. That covers cues like neutral-to-warm expression, eye-line at or just above lens height, modest framing, and stable, even lighting. It penalises chaotic backgrounds, downward chin tilt, low pupil visibility, and over-saturated filters that read as inauthentic. In dating mode the same backbone up-weights approachability and lead magnetism: genuine smile, candid expression, and contextual variety across the slate. The full per-metric breakdown for your face, including the structural inputs that do not change between contexts, lives in the face report once you run a scan.
How a recruiter reads the photo vs how the model scores it
Photo-rating tools that ask strangers to vote on three social traits (competent, likeable, influential) are useful for sentiment but they do not tell you why a photo lost. Aesthetic scorers do the inverse: they break the face into geometry but never connect that geometry to the half-second judgement a hiring manager actually makes. The headshot ranker on this page does both, side by side, so the perception read and the structural read sit on the same screen. The first-impression window here is grounded in Willis & Todorov 2006 (PMID 16866745), which established that trait judgements from faces stabilise within roughly 100 milliseconds of exposure.
How a recruiter scoring at a glance pattern-matches the photo on three social signals.
- Competence72/100
Eye line, framing, posture
- Likeability64/100
Smile genuineness, brow tension
- Confidence78/100
Chin tilt, gaze stability
What the geometric model actually measures, frame by frame, on the same upload.
- Smile68/100
- Skin74/100
- Symmetry81/100
- Jawline70/100
- Harmony76/100
How the two columns connect (illustrative)
Each perception trait is driven by two or three structural inputs. A low Likeability score with a high Smile score usually points at brow tension or eye-line drift; a low Competence score with high Symmetry usually points at framing or chin tilt. The free ranker writes that translation in plain language so you act on the structural cause, not the perception symptom.
That bridge is the entire reason RealSmile separates the headshot mode from the dating mode and from the bare aesthetic face report. A photo with strong structural symmetry but weak Competence is the wrong LinkedIn primary even though it would win a beauty rank, because the recruiter never sees the symmetry — they see the chin tilt and the background. The two-column read above is what every upload returns.
Headshot mistakes that tank LinkedIn engagement
Pattern-matching across thousands of ranked uploads, a small set of mistakes accounts for most of the score gap between a strong professional photo and a weak one. The first is using a cropped group shot. Even with the second face removed in Photoshop, the residual lighting direction, head tilt, and crop ratio almost always read as off. The second is the front-facing webcam photo taken at desk height, which puts the camera below eye line and exaggerates the lower face while compressing the forehead.
The third is heavy filtering. Smoothing the skin to a porcelain finish flattens the cues the model uses to read texture, age, and three-dimensionality, and recruiters consistently rate filtered photos as less trustworthy than mildly imperfect natural shots. The fourth is the “event step-and-repeat” photo with a sponsor logo wall behind your head; the logo competes for attention and your face loses contrast. The fifth is wearing sunglasses or any eyewear with strong reflective glare that hides the eyes. The eye line is the single highest-weighted region in the entire ranker. If a specific metric like ocular asymmetry or eye-line drift keeps showing up as the lowest-scoring axis across your slate, the free face symmetry report breaks the composite into the underlying geometric signals so you can see exactly which feature is dragging the rank.
Lighting, framing, expression — what AI can and cannot fix
A surprising amount of headshot quality is recoverable in software. Uneven exposure can be flattened with a luminance curve. A slight head tilt can be corrected with a level rotation that costs only a few pixels of crop. A noisy background can be replaced or blurred. Skin can be evened out without erasing texture if a light-touch frequency-separation pass is used rather than a global smooth. Colour temperature drift between photos can be neutralised so a slate of headshots looks like one consistent set.
What AI cannot fix is structural. If the camera is below your chin, no amount of post-processing will undo the foreshortening of your forehead. If your eyes are closed, generative fill that opens them never matches the rest of the face under recruiter scrutiny. If the expression reads as forced, with tight lip corners and no orbicularis engagement around the eyes, a smile transplant always lands in the uncanny valley. The honest fix for those three is a reshoot. The ranker on this page tells you which of your existing photos are structurally salvageable and which are not, before you spend a cent on retouching.
One practical rule from the rank data: the highest-scoring professional photos almost always come from a camera positioned at or just above pupil height, roughly six to eight feet away, with a soft directional light source forty-five degrees off-axis. That single configuration solves foreshortening, eye-line, and harsh shadow problems at once. If your existing slate was shot at arm's length under overhead lighting, the ranker will say so, and a ten minute reshoot under window light will out-score any retouching pass on the original.
Pricing tiers for headshot reviews
The free tool on this page is the entry point. It ranks your slate, calls out the lead candidate, and flags the structural problems worth fixing. For most users that is enough to set a primary photo with confidence. When the stakes are higher (an executive search, an inbound founder hunt, a public-facing speaking circuit) the $149 Pro Audit returns a 5-page PDF re-weighted for competence, trust, and Recruiter Confidence — covers all 17 metrics, a ranked recommendation across up to 10 photos, a delete list, and a shoot brief for the gaps the slate does not cover. The $29 Premium Audit (regular $49) is the dating-photo equivalent — same engine but tuned for swipe-rate context.
Compare the full ladder, including the lighter and heavier tiers, on the tools pricing page. If you also want a structural face report (the 17-metric breakdown of jaw angle, canthal tilt, midface ratio, and the rest) the Looksmaxxing Test sits alongside the headshot ranker and uses the same upload. You can run both off a single photo in under a minute by starting with the Pro audit flow.
Decision flow: which photo to set as your LinkedIn primary
The ranker gives you a composite score per photo, but the decision rule for picking the primary is more constrained than “take the highest number”. A LinkedIn primary sits in three places at once: the profile hero, the small thumbnail in the connection feed, and the avatar circle in messages. A photo that scores well in one of those three contexts but falls apart in another is the wrong primary, even if its overall composite score is the highest of the slate. The decision flow below walks through the four-step rule the ranker applies before naming a winner.
Step one is the small-thumbnail test. The model crops every photo to a 64-pixel square and re-scores facial-feature legibility. Photos that score well at full size but lose the eye line at thumbnail are downgraded, since most LinkedIn discovery happens through small thumbnails in feed and search. Step two is the half-second attention test, which simulates the Princeton/Todorov 2006 first-impression window by re-rating warmth and trust at a 100-millisecond exposure budget rather than the longer rater task used for the full composite. Photos that depend on a slow read (a complicated background, a subtle expression, a non-obvious primary subject) lose ground at this step and rise back only if the long-read score remains anchored to a clean fast-read.
Step three is the search-result triangulation pass. The ranker checks whether a recruiter who lands on your profile from a search result built on a different keyword (your job title, your company, your university) would still recognize the photo as belonging to a person matching the rest of the page. A photo of you in a wetsuit on a sailboat is a strong dating-app candidate and a weak professional primary because the wetsuit is the loudest element in the frame. Step four is the slate-redundancy check. If your three highest-scoring photos all show the same angle and lighting, the ranker recommends keeping only the strongest of the three as primary and rotating the other two into the activity timeline so the slate as a whole reads as a real human rather than a single repeated render. The four-step verdict is summarized at the top of your AI face report card in plain language so the primary-photo decision lands as a single recommendation rather than a wall of raw scores.
The single biggest mistake the rank data surfaces is overwriting a strong primary with a fresh upload simply because the new shot is more recent. Recency does not beat composite score on its own. If the new photo loses any of the four steps above, the ranker recommends keeping the old primary and adding the new one as a secondary slate photo until the next refresh cycle. The free tool on this page runs all four steps automatically and writes the verdict in plain language so you can act on it without reading raw scores.
Headshot context matters: 4 use cases, 4 different optima
The same face benefits from different framing depending on where the headshot will be used. Below are 4 common contexts and what each rewards.
LinkedIn / Corporate
Top trait priority
Trustworthy + Competent
Visual cue
Business-formal collar, neutral solid background, even front-light, slight head tilt.
Common pitfall
Outdated suit + harsh fluorescent ceiling light.
Real-estate / Sales
Top trait priority
Approachable + Confident
Visual cue
Warm Duchenne smile, direct eye contact, branded color background, mid-shot crop.
Common pitfall
Overly formal stiff pose that reads cold to first-time buyers.
Tech founder / Author / Speaker
Top trait priority
Credible + Visionary
Visual cue
Candid mid-action shot, environmental detail (workspace/stage), softer styling, off-center framing.
Common pitfall
Generic studio backdrop that reads as recruiter-photo, not personality.
Dating profile primary
Top trait priority
Warm + Authentic
Visual cue
Outdoor soft natural light, genuine smile, subtle context (cafe, outdoors, hobby), eye-level lens.
Common pitfall
Corporate LinkedIn headshot reused as dating-app primary — reads as cold and effortful.
Use-case framings are heuristics derived from common professional photography practice and dating-photo research; individual industries and platforms vary.
Grading rubric — by where the photo is going
Same face, different rubric. A headshot that scores 92 for LinkedIn might score 64 for a dating profile because the signals are weighted differently. Use the table below to grade your photo against the rubric that actually matches your use-case.
Top of rubric
Eye contact + warmth-first smile
Middle band
Neutral background, business-appropriate wardrobe, clean grooming
What disqualifies
Avoids: full-body crops, party photos, sunglasses, filters that smooth skin texture
Weight rationale
Trust > attractiveness. A LinkedIn recruiter screens for "would I hire this person" — symmetry and dominance matter less than approachability.
Top of rubric
Range — emotional read on the face, eye life, micro-expression
Middle band
Plain background, no props, minimal post-processing, type-clarity (does the face read consistent with castable archetype?)
What disqualifies
Avoids: glamour retouching, busy backgrounds, anything that obscures the face from cheekbones up
Weight rationale
Castability > attractiveness. The casting director is sorting by archetype fit, not by hotness — a clearly readable type beats a "pretty" but ambiguous photo.
Top of rubric
Approachability + confidence balance + clear face visibility under 0.5s
Middle band
Authentic context cues (where you are, what you do), genuine smile or relaxed neutral, eye-line slightly above lens
What disqualifies
Avoids: corporate stiffness, group photos as the lead, low-resolution, mirror selfies, gym mirror flexing
Weight rationale
Pass-through > polish. The job is to survive the 0.5-1s thumbnail scan and earn a profile open. Studio polish often reads as "online photographer" and reduces match rate.
Top of rubric
Authority cues + eye-line at lens or slightly below + intentional posture
Middle band
Slight contextual background (stage, studio, brand color block), wardrobe consistent with positioning
What disqualifies
Avoids: stock-photo neutrality (looks like a template), heavy retouching that loses skin texture
Weight rationale
Recognition > generic appeal. The photo runs on conference programs, podcast tiles, and press releases — needs to be the face the audience already remembers seeing.
Top of rubric
Region-specific norm match (US: usually no photo; EU/Asia: high norm-fit weight)
Middle band
Neutral expression or light smile, business attire, lighting that flatters without distracting
What disqualifies
Avoids: outdated photo (older than 2 years), inconsistent with LinkedIn, anything cropped from a casual shot
Weight rationale
Norm-fit > distinctiveness. The recruiter compares your photo to a stack of similar applicants — looking different in this format is usually a negative signal, not a positive one.
The same photo can be a 92 for LinkedIn and a 64 for Hinge. Pick the rubric your photo is competing under before deciding whether the photo is "good."
Frequently asked questions
How long does an AI headshot review take?+
The free in-browser ranker returns a result within roughly 8 to 20 seconds per photo, depending on your device. The Premium Audit ($29 launch, regular $49) is delivered as a 5-page PDF, typically within a minute of upload. Nothing queues, nothing schedules — the model runs the moment you submit.
What image format and size should I upload?+
JPEG, PNG, HEIC, and WEBP are all supported. We recommend at least 1024 pixels on the long edge so the model can read pupil position, jaw outline, and skin texture cleanly. Files larger than 12 MB are downscaled in your browser before any network round trip — your original is never sent.
Can I upload multiple headshots in one session?+
Yes. The free LinkedIn ranker accepts 2 to 6 photos and ranks them head-to-head. The Premium Audit accepts up to 10 photos and identifies which one to set as your primary, which to keep as a secondary, and which to remove from your profile entirely.
Will my photos be stored or used to train any model?+
No. Photos analyzed in the browser never leave your device. Photos that route through our mobile pipeline are deleted from disk immediately after the score returns; nothing is retained, indexed, or used as training data. The privacy badge at the top of every page links to the policy.
Is this the same model used for the dating photo audit?+
The geometric scoring stack (symmetry, expression, framing, lighting) is shared. The weighting is different. Headshot mode prioritizes perceived trust, competence, and professional warmth; the dating mode prioritizes approachability, attractiveness, and lead-photo selection. Same engine, different objective function.
How should the LinkedIn primary photo differ from a CV photo or a Slack avatar?+
The LinkedIn primary is a half-bust frame at pupil height, framed loose enough that crop is breathing room rather than a tight head-and-shoulders crush. A CV or resume photo (where it is still expected, mostly outside the United States) sits closer to a passport crop with a neutral light background and no environmental cues. A Slack or Teams avatar is a cropped square of the LinkedIn primary, usually pushed a bit tighter so the face still reads at 32 pixels in a sidebar. The headshot ranker scores the same slate against all three contexts and flags any photo that scores well as a primary but falls apart at small avatar sizes due to facial-feature legibility loss.
Should I use the same headshot across LinkedIn, my company bio, and the speaker page for an event?+
For brand recognition the answer is yes within a single hiring season, then refresh as a slate every twelve to eighteen months. Recruiters and inbound prospects pattern-match on the photo as part of the name; switching mid-funnel costs more than a slightly older shot. The exception is when one venue requires a markedly different aspect ratio or background. Speaker bureaus often want a darker editorial background for stage banners, while LinkedIn rewards lighter, neutral backdrops for thumbnail contrast. The ranker handles both cases by scoring against context-specific weights and recommending which photo to submit where.
How often should I refresh my professional headshot?+
A practical rule from the rank data is every twelve to eighteen months, or sooner whenever a visible change in glasses, facial hair, hairstyle, or weight makes the current photo read as outdated. Recruiters who interview a candidate in person and then look at a year-old LinkedIn photo report a small but measurable trust hit when the gap is large. The free ranker on this page is intentionally fast so that an annual reshoot can be tested against the existing primary in under five minutes, with the new photo replacing the old one only if it scores cleanly higher across the trust and competence axes.