Increasing profile quality by improving the photo upload flow

Kismia is a dating app, profile pictures directly impact engagement and monetization. Most users had only 1 photo, reducing profile quality and platform trust, which hurt match rates and retention.

I redesigned the photo upload flow in the onboarding funnel, increasing the average number of photos per user, in collaboration with PM and data analyst.

+38%

Average photos per user

+19pp

Users with 2+ photos

+4.5pp

At least one uploaded photo

Role: Product Designer

Team: PM, Analyst, Engineer

Duration: 4 weeks

Type: A/B test (66k users)

Initial Analysis

The flow reduced signup drop-off by allowing users to skip steps after adding their first photo. This made it easier to complete the funnel but didn’t encourage uploading multiple photos. Fewer photos lead to lower match rates, weaker retention, and reduced monetization.

75%

Uploaded only one photo

64%

Users with one approved photo

40%

Users with two verified photo

Goal

Increase the number of users uploading 2+ photos while maintaining onboarding completion rates.

Hypothesis 1: Users don’t know what photos to upload

Users don’t clearly understand what types of photos are acceptable. As a result, they upload low-quality photos that fail moderation, which discourages them from adding more.

What I did

Interviewed the photo moderation team to identify the most common mistakes in uploaded photos

Designed a popup with visual photo tips and examples based on moderator feedback

Tested the solution via A/B experiment against the control flow

Outcome

No statistically significant improvement

The number of uploaded photos did not increase

Key Insight:

Users still don’t know what photos to upload or aren’t motivated to upload more after an initial attempt, even with guidance.

Hypothesis 2: Upload flow reduce motivation to add more photos

Users need a clearer completion goal and stronger motivation, not instructions.

What I did

Benchmarked leading dating apps to identify effective contribution patterns

Identified that visible empty photo slots drive progression and increase upload depth

Shifted the focus from educating users about photo quality to motivating completion through clear structure

Aligned the approach with PM

Results

+38%

Average uploaded photos per user

+19pp

Users with 2+ approved photos

+4.5pp

At least one uploaded photo

Key Learnings

The main impact didn’t come from convincing inactive users to upload a photo.
It’s often more effective to support users who already show intent than to push those who are inactive.

If the flow feels simple and valuable, users are more likely to continue and contribute more.

© 2024 Illia Redunenko