Try-on filters were, for a long time, a marketing novelty — a fun way to get someone to open an app, take a selfie with a lipstick shade, and post it. The conversion numbers rarely justified the engineering effort. That's shifted, and the shift is worth paying attention to because it says something about where consumer AR is actually finding a business model, as opposed to just an audience.
The unglamorous problem AR solved
Online beauty and fashion retail has always had one structural weakness: you can't know how something will look on you until it arrives. Return rates for color cosmetics and eyewear bought online have historically been brutal, and returns are expensive in ways that aren't obvious from the outside — restocking, shipping both ways, and the margin hit on anything that can't be resold.
Accurate face and body tracking closed a meaningful chunk of that gap. Shade-matching that accounts for actual lighting conditions, frame try-on that reads your face shape instead of just overlaying a flat image — these aren't flashy features, but they're the ones that move a return rate from painful to tolerable.
Where it's heading next
The next frontier isn't better tracking, it's better data feeding the tracking — brands are starting to pair try-on with actual undertone and skin-type detection, so the recommendation isn't just a visual match but an explanation of why a shade works for you. That's a genuinely useful step past novelty, and it's the kind of feature that keeps people using try-on tools after the first week of curiosity fades.
The honest caveat: try-on still struggles with texture and finish — matte versus dewy, structured fabric versus something that drapes. Getting that right is less about tracking and more about rendering, and it's the gap most retail AR still hasn't closed.
