Converting Session Data into Conversion Fixes with AI

Executive Summary

This is not a redesign. This is a UX audit using MS Clarity session data and Claude to identify where and why users were dropping off in Jolt Health's multistep onboarding quiz.

Discovery

I analysed MS Clarity session data and mobile heatmap to identify friction patterns across the onboarding flow. Behavioural observations were then fed into Claude to synthesise findings and surface prioritised design recommendations.

Design Direction

1. CTA Invisibility at Quiz Entry

Mobile users tapped the Free Assessment label and non-interactive hero elements before locating the Get Started CTA. The Free Assessment label reads as a primary action, pulling attention away from the actual CTA.

Recommendation: Reduce the visual weight of the Free Assessment label so it does not compete with the primary CTA. Ensure Get Started is the only element that reads as a dominant interactive action on the page.

Design Direction

1. CTA Invisibility at Quiz Entry

2. Scroll Burden at Date of Birth Locks Out Older Users

The date of birth picker requires users to scroll through a drum-style selector. For users born in the 1950s or earlier, reaching their birth year requires dozens of scroll interactions — creating significant friction at a mandatory input stage.

Recommendation: Replace the drum picker with three separate select dropdowns for Day, Month, and Year. Allow direct year input so users can type their birth year rather than scrolling to it.

Design Direction

1. CTA Invisibility at Quiz Entry

Results

Outcomes

This audit produced a prioritised fix list:

Fix now. Reduce the Free Assessment label visual weight and replace the DOB drum picker. Both are causing abandonment before sign-up.
Next sprint. Redesign mobile hero CTA for visual dominance and introduce direct year input on the date picker.

Two A/B experiments were proposed to validate key recommendations with real traffic data.

View full audit

Reflections

I fed behavioural observations into Claude and received a structured audit, complete with issue severity classification, UX principles, fix recommendations, and A/B test suggestions, in minutes. What would have taken hours of manual synthesis was done faster and with more rigour. This workflow is now part of how I approach any UX audit.

Back