Sajjad Abedi
Users were failing verification because of poor photo quality and confusing processes. I redesigned the capture experience with real-time feedback and ML assistance, reducing resubmissions by 23% and fraud by 25% while improving overall success rates.

Veriff auto capture

Tl;dr

Users were failing verification because of poor photo quality and confusing processes. I redesigned the capture experience with real-time feedback and ML assistance, reducing resubmissions by 23% and fraud by 25% while improving overall success rates.

The challenge

Users were abandoning Veriff's verification process at alarming rates. High resubmission rates were costing us money, frustrating customers, and causing us to lose business to competitors. Despite going through company changes and layoffs, our team had to solve this problem quickly.

The root cause

Users were failing verification for two main reasons: poor image quality and confusion about the process. But I needed data to prove this and find specific solutions.

Research & discovery

I started by analyzing our data to understand the scope of the problem. I looked at failure rates by country, customer type, and verification step. The data showed clear patterns, but I needed to understand why users were struggling.

Deep dive into the data.
Deep dive into the data.

Next, I dug into customer feedback. I reviewed support tickets, listened to customer calls, and shadowed our verification team. The patterns were clear: users submitted blurry photos and didn't understand what was expected of them.

Shadowing the manual verification team.
Shadowing the manual verification team.

A quick wins

I started by helping our verification team work more efficiently. We added machine learning tools that highlighted potential issues, helping reviewers catch problems faster. This reduced human error and showed that technology could help our team instead of replacing them.

Verification tools
We brough the ML outcome to the manual verification
We brough the ML outcome to the manual verification

Image quality problem

The bigger challenge was helping users take better photos. I designed a new capture experience that gave users real-time feedback on lighting, framing, and image quality. Instead of letting users submit poor photos, we helped them get it right the first time.

Starting with welcome screen, and making sure the user is ready to capture the image.
Starting with welcome screen, and making sure the user is ready to capture the image.
Automatically deteching the document type and capturing the image.
Automatically deteching the document type and capturing the image.
Automatically capturing the selfie image.
Automatically capturing the selfie image.
Giving feedback to user to make sure the image is good enough.
Giving feedback to user to make sure the image is good enough.

We tested these solutions with real users and refined them based on feedback. The key was finding the right balance between being helpful and not overwhelming users with too much guidance.

The impact

The results exceeded our expectations:

This project proved that combining data insights with user-centered design could solve complex business problems while improving the customer experience.