The AXA result challenged the proposed pattern
The AXA mobile comparison project tackled a familiar problem: insurance formulas, packs and options need an overview, yet a phone cannot comfortably display everything side by side. The proposal was a swipable matrix, developed as a running Figma prototype.
User testing found a preference for the existing card-based version. The work redirected the design toward a second iteration for further testing. The record does not publish the participant count, test tasks, detailed reasons for preference or a completed V2 result.
The defensible lesson is therefore about the decision process: the prototype made an alternative testable, and the test prevented treating that alternative as a proven improvement. It does not establish that cards always win, or that horizontal comparison always fails.
Decide what people need to compare
Before choosing cards or a matrix, define the comparison task with the product team. Is someone deciding between coverage levels, checking a specific condition or discovering what an optional pack adds? Use consistent language and ensure each displayed value refers to the same kind of information.
For a new design, I would investigate where people lose context. Can they still identify the formula after moving through details? Can they find the explanation behind a shorthand label? Do they understand a blank cell, an unavailable option and information that has not yet been provided as different states?
These are proposed review questions rather than findings from the AXA test. Their purpose is to expose a practical risk before selecting an interaction based on visual neatness.
Ask for a decision and its reasoning
Compare the current interface with the candidate using the same realistic tasks. Ask participants to find a difference that matters to their scenario, explain it in their own words and revisit an option after inspecting another. Record incorrect interpretations and requests for help alongside task completion.
Preference is useful evidence, but separate it from demonstrated comprehension. Document devices, languages, participant characteristics and the scope of the test before describing a winner. Where evidence remains mixed, keep the uncertainty visible and define the next iteration around the unresolved task. That follows the useful part of the AXA example: allowing the research to change the design direction.
Your brief
Tick what you can already answer. Unknowns are fine: just name them.
Questions & answers
Are cards better than tables on mobile?
There is no winner established here for every product. AXA participants preferred the existing cards in one exploration. Your decision needs the actual comparison task, content and user testing.
Did the AXA matrix improve conversion?
No conversion improvement is documented for the matrix. It was a prototype that informed a second iteration after users preferred the old version.
What should a mobile comparison test measure?
Check whether people find a relevant difference, explain it correctly and retain context while moving between options. Include navigation difficulties and preference, with the sample and test limits recorded.
Sources & context
- AXA: Mobile Comparison Matrix (2024)
Documents a Figma exploration, preference for the existing cards in user testing and a second iteration awaiting further validation. It does not establish successful rollout. First-person project account; the underlying research and analytics are not published here. Checked .
- W3C: Understanding WCAG 2.2, Target Size (Minimum)
Primary explanatory guidance fetched on 10 November 2026. Describes SC 2.5.8 and its exceptions; this single criterion is not a complete accessibility assessment or a legal opinion. Checked .
Test your mobile product comparison
Share the options people need to compare and the current mobile experience. We can define representative tasks and review a focused alternative.