Fifty-fifty overall. Until we looked at who was actually buying.
PistonHeads was trapped in an expensive product stalemate. Standard A/B testing on a core search UX decision - whether users should search by single models or multiple models - yielded completely inconclusive data. The internal team was deadlocked, unable to move forward on a business-critical feature because the quantitative metrics provided no clear direction.
Designed a survey that separated stated preference from revealed behaviour, cross-referencing responses with site usage and sales data to profile the highest-value user segment.
The A/B test failed because it treated all traffic as a single monolith. The research revealed a stark commercial split: users demanding single-model selection possessed significantly higher site experience, stronger purchase intent, and vastly superior lifetime commercial value. By optimizing for this high-value behavioral cohort rather than the "average" user, the product team secured a clear, defensible roadmap backed by financial reality, not internal opinion.
