Why did Umax give me a low score? A calmer, more accurate answer
Umax gave you a low score and it stung. Here's what that number actually measured — and why a single geometry score isn't your face's real-world appeal.
You took the photo. You waited through the scan that froze somewhere around 91%. And then a number came back — lower than you hoped, lower than the before-and-after clips in the ads — and it sat there in your chest for the rest of the day.
First thing, before any explanation: that number is not a verdict on your face, your dating prospects, or your worth. It feels like one. It is not one. Stay with me and I'll show you why, mechanically, not as a pep talk.
Because here is the part nobody leads with. The people who built these apps know the score moves you. That's the product. A calm, accurate number that said "you're fine, go live your life" would not sell a $3.99/week subscription. The discomfort is doing a job. Once you see the machinery, it loses most of its grip.
What the Umax score actually measured
Umax — a face-scan app from Blake Anderson, reported by Fortune at 7M+ downloads — takes one photo and runs facial geometry on it. Jawline angle. Cheekbone projection. Symmetry. A "masculinity" estimate. It maps those measurements onto a 0-100-style number and, in the looksmaxxing world, often a PSL label on top.
Read that again slowly. It measured the geometry of one static image. Not your face — that image. Not how you come across — the pixels you happened to feed it.
That distinction is the whole ballgame, and it explains the single loudest complaint about these apps: the score won't hold still. Users on the App Store describe submitting the same picture more than once and getting a different number each time — one reviewer's phrasing was "submitted the same picture 3 times, got a different number." Others call it "completely inaccurate" or say the result didn't match the TikTok ad. I'm quoting those users, not asserting it myself — but the pattern is consistent enough to mean something.
Caveat: facial geometry is real and measurable. The problem isn't that Umax measures nothing. It's what the measurement is, and isn't, connected to.
Key numbers
- Umax analyzes a single static photo; users report the same image returning different scores on repeat scans.
- A 0-100 geometry score is not calibrated against any dataset of real-world attraction outcomes.
- Across 919 studies and 12,261 raters, strangers agree on facial attractiveness ratings — but those are lab ratings, not dating decisions (Langlois et al., 2000).
- Lighting, angle, and lens alone can swing perceived attractiveness 1-2 bands with zero change to your bone structure.
- A first impression forms in roughly 100 milliseconds — and it's running on expression and approachability, not a jaw-angle measurement (Willis & Todorov, 2006).
Why the same photo gives different scores
This is the tell, so let's sit on it.
If a measurement of a fixed object changes every time you take it, the measurement is unstable — not the object. Your jaw did not move between scan one and scan three. What moved was the model's read of a few thousand pixels, and those pixels are absurdly sensitive to things that have nothing to do with you.
Tilt your chin down two degrees and the jaw angle the app "sees" changes. Shoot under flat overhead light and your cheekbones flatten; shoot near a window and they pop. A phone's wide front camera bows your nose and widens your face the closer you hold it. None of this touched your actual face. All of it moves the number.
So when the score comes back low, the honest reading is not "my face scored low." It's "this particular photo, under this particular light, at this particular angle, through this particular lens, produced a low geometry read today." That is a much smaller, much more fixable thing.
Caveat: a brutally bad photo of a genuinely photogenic person still exists. The point isn't that photos lie equally in all directions — it's that a single one is far too noisy to rule anything about you.
The calibration gap nobody mentions
Here's the deeper problem, and it's the one that should actually take the weight off.
Suppose the geometry score were perfectly stable. It still wouldn't tell you what you came to find out — whether people are drawn to you — because it was never checked against that outcome. A 0-100 face score is not validated against any dataset of who actually got approached, matched, or asked out. It's a number generated by a model trained to output numbers. There's no real-world attraction data on the other end of it confirming the score means what you assume it means.
This is the quiet difference between a measurement and a fortune. A bathroom scale is calibrated against actual mass. A looksmaxxing score is calibrated against… other looksmaxxing scores, mostly, and an aesthetic the subculture decided on. When it tells you "62" or "PSL 4," there's no validation step underneath confirming that number predicts a single real interaction in your real life.
That's not a knock on the math. It's just what the math is — and isn't.
Caveat: validation is hard and genuinely expensive, which is partly why almost nobody in this space does it. We're not claiming a perfect instrument exists. We're saying the absence of one should make you hold any single score very loosely.
What actually drives whether people are drawn to you
Now the part that should change how you feel, because this is where the research goes somewhere kinder and more honest than a jaw-angle number.
Yes, strangers broadly agree on facial attractiveness in a lab — that's the Langlois et al. (2000) meta-analysis, 919 studies, real and worth reading. But "agreement in a lab" is not the thing you actually live inside. The thing you live inside is the first 100 milliseconds of a real encounter, and Willis and Todorov (2006) showed that in that window people are forming snap judgments of trustworthiness and approachability — reading your expression and bearing, not pulling out a protractor for your cheekbones.
That window runs on signals a still-photo geometry scan can't even see:
- Expression. A relaxed, genuinely warm face reads as more attractive than a tense symmetrical one. The eyes and mouth do enormous work here, and Umax's geometry pass largely ignores them.
- Approachability. Whether someone reads as warm and open is a first-impression driver in its own right — and it's almost entirely about bearing, not bone (Todorov's first-impressions work).
- The halo around all of it. People consistently assume good-looking and warm-and-confident people have other good traits too (the "what is beautiful is good" effect, Dion et al., 1972) — and warmth and confidence are things you can actually move.
None of that fits in a 0-100 from one selfie. Which means a low Umax score isn't measuring most of what determines whether someone wants to talk to you. It measured a sliver, on a bad day, with a noisy ruler.
Caveat: none of this means looks "don't matter." They clearly do. It means the part that matters is broader, more dynamic, and far more in your hands than a single static geometry score implies.
If the score got into your head
This is the part I most want you to hear, because the looksmaxxing space has a real cost here. Psychologists quoted by Fortune have warned that these rating apps are feeding a male body-image problem — Mount Sinai researchers specifically flagged the risk of an algorithm delivering a stream of negative appearance feedback to young men, and the body dysmorphia that can follow.
If a number sent you spiraling, that is not a character flaw and it is not unusual. It's the predictable result of handing your self-image to a slot machine that profits from your unease.
So, concretely: a single app score is not data about your life. The same photo giving three answers is proof the instrument is shaky, not proof you're "really" the low one. And the parts of attractiveness that actually move outcomes — your expression, your warmth, your bearing, your photos, your fitness, your style — are exactly the parts a geometry scan can't see and you can absolutely change. That's the hopeful read, and it's also the accurate one. Those two usually aren't the same thing. Here they are.
For the longer version of why a stable, calibrated geometry score still wouldn't be the answer, PSL scores vs. objective beauty is the deep dive. And what women actually find attractive walks through the drivers above in detail.
A different kind of read
If you want a picture of how you come across — not a verdict, a read — that's the gap we built the test to fill. It doesn't crown you a number out of 100. It looks at how a real first impression forms: the expression and approachability you're projecting, who's most likely drawn to you, and the specific, movable things that would shift it. No paywall to see your result, and nothing about it is designed to make you feel worse so you'll pay to feel better.
A few honest places to go from here. How lighting and angle alone move your perceived score covers exactly why one photo is so unreliable. The first-impression window covers what's actually happening in those first milliseconds. And if you specifically want to compare what a looksmaxxing number says versus how a face lands in real life, Umax score vs. real life takes that head-on.
You can also just close the app for a week. The number was never measuring the thing you were worried about. Your face on a good-light day, relaxed, talking to someone who likes you — that's the version that's real. The scan never met that guy.
Studies referenced: Langlois, J. H., Kalakanis, L., Rubenstein, A. J., Larson, A., Hallam, M., & Smoot, M. (2000). Maxims or myths of beauty? A meta-analytic and theoretical review. Psychological Bulletin, 126(3), 390-423. Willis, J., & Todorov, A. (2006). First impressions: Making up your mind after a 100-ms exposure to a face. Psychological Science, 17(7), 592-598. Dion, K., Berscheid, E., & Walster, E. (1972). What is beautiful is good. Journal of Personality and Social Psychology, 24(3), 285-290. Reporting on Umax and looksmaxxing apps: Sternlicht, A. (2024). Fortune.