// Case study / Experimentation
The quiz worked. The question was where to put it.
A B2B/B2C hardware authentication brand had a product selection quiz quietly driving purchase intent — but it was under-surfaced. Phase 1 used GA instrumentation to confirm the conversion signal. Phase 2 used VWO to A/B test four store-page placement variants. Quiz-exposed users outconverted non-exposed users across both phases; the optimal placement was determined by measurement, not assumption.
What happened
Instrumentation confirmed the signal. A/B testing found the surface.
A hardware authentication brand sold a range of products differing by encryption method and hardware connector type. A product selection quiz existed to help users navigate the complexity — but it was not prominently placed. Phase 1 GA analysis confirmed that quiz-exposed users consistently outconverted non-exposed users across sessions; non-exposed sessions declined over the same period. Phase 2 A/B tested four placement variants on the main store page. The prominent entry-point variant generated roughly twice the pathed conversion rate of the control and outperformed all alternatives.
Context
Wide product range, high-consideration purchase, no clear entry path.
The client sold hardware security keys across a range of form factors and connectivity standards. For an entry-level buyer without an enterprise IT background, the catalogue was genuinely hard to navigate. A product selection quiz had been built to address this friction, but it was embedded in the flow without a prominent entry point from the store page, where the majority of purchase-intent traffic landed.
The initial brief was general conversion optimisation. What changed the shape of the engagement was data: Phase 1 GA analysis surfaced the quiz’s conversion impact as the more durable signal.
Task
Confirm the quiz hypothesis with data before redesigning around it.
Analytics instrumentation and insights: tracking the quiz flow and conversion pages in GA, analysing the behavioural split between quiz-exposed and non-exposed sessions, implementing VWO for Phase 2, and interpreting the results. The UX team designed the visual variants; the measurement layer and test framing were the analytics contribution.
AControlno quiz bar — baseline
BThick bar~2× control — winner
CThin / subtle> control, < B
DThin / bright≈ variant C
Store page without a quiz entry point. Established the baseline conversion rate for non-quiz-exposed sessions.
Prominent entry banner sized to a hero image. Generated roughly twice the pathed CVR of control and outperformed all other variants across new and returning user cohorts.
Lighter-weight entry point with a muted background. Outperformed control but meaningfully underperformed the prominent variant.
Higher-contrast thin bar. Performed similarly to variant C; both confirmed that reducing the quiz’s visual weight reduced its conversion impact.
Instrumentation note
VWO managed the test; GA tracked the downstream path.
Phase 2 used VWO to allocate users to variants. Custom GA dimensions captured test assignment against conversion events — pathed conversion rate analysis for each variant, not just click-through, segmented by new and returning visitor cohorts.
Outcome
Measurement confirmed the hypothesis at each phase.
- Phase 1 confirmed purchase-intent signal: quiz-exposed sessions converted at a meaningfully higher rate than non-exposed sessions, which declined over the same period.
- Phase 2 confirmed placement as the lever: the prominent thick-bar variant generated roughly twice the pathed CVR of control, alongside a strong revenue lift for quiz-exposed users.
- Subtlety was counterproductive: both thinner variants underperformed the prominent entry point, confirming that a tool designed to resolve product complexity needed to be impossible to overlook.
- Both cohorts responded: Phase 2 lifted conversion for new and returning users, suggesting the quiz had utility beyond first-time purchase decisions.
Design lesson
When product complexity is the friction, a configurator is only useful if users encounter it.
The quiz was working as designed — helping users identify the right product for their use case — but reaching only a fraction of the relevant audience. Measurement identified the gap before further design work was committed. The A/B test made the placement decision in evidence rather than aesthetic preference.
The transferable pattern: when a product line creates genuine selection friction, the measurement question is not “does a configurator help” but “where in the journey does it need to be visible to change behaviour.”