A deep-dive analysis revealed a key issue: 22% of users were failing the photo verification. That meant nearly one in five customers couldn’t complete their identity checks and purchase insurance products.
Here’s what was going wrong:- Poor capture conditions: Users often took photos in motion — inside cars, building lobbies, or outdoors in poor lighting with unstable internet connections. This led to low-quality facial images.
- No feedback loop: The system didn't guide users on how to take better photos or explain why their submission was rejected.
These weren’t fraud attempts — they were real customers trying to complete the remote identification. Yet the used system wasn’t built to handle such real-world cases.
In short, the integrated facial authentication solution lacked flexibility and failed under typical conditions of everyday mobile usage.