How an AI skincare app used real users to find real bugs on the Toloka Platform

The client is an AI skincare app that scans your face, evaluates cosmetic products, and generates personalized routine recommendations. The face scanner is central to the experience. That also made it the highest risk surface to ship with undetected issues.

To stress test the mobile build ahead of a release, the team turned to Toloka for structured real world QA.

The challenge: QA that reflects real-world use

Automated testing doesn't catch what real users encounter. The team needed fresh installs on real devices, with real people going through onboarding the way a first time user would, and a reliable record of what happened when something went wrong.

Toloka helped the team set up a task workflow on our Platform where contributors installed the app from an Expo build link, completed the full activation flow (onboarding, face scan, home screen, App Info capture), and submitted a full screen recording of each session. Each recording was evaluated against four criteria: flow completeness, face scanner completion, screen visibility, and observable app behavior.

What they found

The project moved fast: 50 minutes from setup start to project launch, and less than a day to get every test scenario completed and labelled. More than 80,000 people were invited to participate, and contributors surfaced several application crashes and bugs that hadn't been caught in internal testing. Each test scenario took about 15 minutes to complete. The structured recording format made every issue reproducible, giving the team a timestamped record to work from instead of a written bug report open to interpretation.

Why it worked

Requiring contributors to start recording before opening the app captured edge cases that would have been invisible in a standard bug report, including behavior during connectivity interruptions mid scan. An AI assisted review layer pre-evaluated each session across all four criteria before a human reviewer touched it, so the team spent its time on confirmed issues rather than sorting through raw recordings.

Shipping a mobile app? Find the bugs your internal testing missed.

Run structured QA with real users on Toloka's self service platform. No lengthy onboarding, no minimum spend. This team went from setup to launch in under an hour, had every scenario completed and labelled in under a day, and had a pool of tens of thousands of contributors ready to test. Set up your first task and get results fast.