Frame Fit Without a Selfie
What you’re dealing with
For online eyewear, frame fit is your single largest return driver. Customers don’t know their face width, bridge size, or whether a temple length will pinch behind the ear — so they bracket-buy, and the ones who don’t bracket tend to return. Virtual try-on solves the aesthetic question (“does it look like me?”) but not the fit question (“does it sit right?”) — those are different problems, and the fit one is what costs you margin.
For VR/AR, the physics are harder. A strap that’s 10mm too tight causes pressure headaches. A light seal that doesn’t match face depth leaks light, breaks immersion, and generates the most specific returns reason in your dataset. And unlike eyewear, you can’t really bracket-buy a $500 headset.
Where the money is
- Eyewear returns: Frame width mismatch is the plurality cause of online optical returns. Pre-routing to the right frame front width and bridge size before dispatch removes the most common return reason from your funnel.
- VR/AR accessories: Correctly-sized straps and light seals sold as SKU variants (rather than a universal “one size”) reduce the “fits poorly” return category and open a legitimate accessory attach path.
- Onboarding: Two numbers most customers already know — height and weight — captured at the point of size selection. Customers who share them in exchange for a confident fit recommendation are the same customers who would otherwise have bracketed or returned.
What we actually do
We predict up to 24 craniofacial dimensions from height and weight — numbers your customer knows by heart and will share when there’s a clear reason to. Bizygomatic (cheekbone) breadth for frame front width. Nasal root breadth for bridge geometry. Head circumference for headset strap sizing. Menton-sellion length (chin to nose bridge) for light-seal selection. No photos, no LiDAR, no extra checkout step.
Regional head shape is real and we handle it. Head breadth, nasal bridge, and bizygomatic breadth differ measurably between East Asian, European, and other populations. Our 7-region calibration routes an Asian-fit variant vs. standard variant automatically — which is the thing that makes this scalable across markets without per-region fit studies.
Where we’re upfront
Craniofacial dimensions are soft-tissue influenced and sit at slightly lower confidence than skeletal body dimensions — typically 70–80 on our index with height+weight anchors. That’s the right precision for routing a customer to the correct frame width SKU or headset strap size; it isn’t a substitute for a dispensing optician’s PD measurement on prescription lenses, and we don’t position it that way.
For everything else — which is the majority of your returns — here’s what it means: the customer trades two numbers they already know for a confident size recommendation, the frame front width and bridge are right the first time, and the return reason you see most often in your dashboard stops appearing as often. Same checkout, fewer parcels coming back.
Farkas, L.G. et al., “International Anthropometric Study of Facial Morphology in Various Ethnic Groups,” Plastic & Reconstructive Surgery, 2005. Population calibration: ANSUR II (North America) + regional anthropometric sources; coverage notes in API reference.