How a quiz mechanic works

A quiz is a structured dialogue between the shopper and the site: a series of questions answered by
a click or a short text entry. From the answers, the system assembles a personal product selection.

A classic quiz runs three to seven questions. More and drop-off rises; fewer and there is not enough
to select on. The right length depends on the category: five or six questions are justified for a
mattress, two or three for a phone case.

Tip: start with the most important question rather than the easiest one. Someone who has
answered a meaningful question is already invested and more likely to finish.

The double benefit

For the shopper: a quiz removes choice paralysis on a wide catalogue. Instead of working through
500 items, they get a personal shortlist of five to ten.

For the business: a quiz collects zero-party data — explicitly stated preferences. Unlike
behavioural data, which records what was viewed, zero-party data states intent: looking for a gift,
this budget, this ingredient matters.

Quizzes by vertical

Vertical Typical questions Value
Fashion Style, size, occasion, budget Outfit selection, sizing
Beauty Skin type, concern, sensitivity Avoiding reactions, finding a fit
Electronics Who for, the task, budget, experience Removing technical anxiety
Furniture / DIY Room size, style, material Fitting a specific space

Quiz or AI dialogue

A classic quiz is a fixed decision tree with predefined answer options. The modern alternative is
conversational discovery on an LLM: the shopper writes freely, the system interprets the request and
clarifies parameters in dialogue. The LLM route is more flexible and demands more infrastructure —
retrieval over a catalogue in a vector database.

Placement and measurement

Placement: homepage, category page, product page (a help me choose widget), exit-intent popup,
mobile app onboarding.

Key metrics: completion rate, click-through from results into a product page, add-to-cart rate
among quiz takers, and quiz against control in an A/B test.