Case study
Helping people who know the person, not the product.
Context
Gift shops sell to people buying for someone else. In the shop, a good seller asks two or three questions and puts the right object in their hands.
The problem
Online, the visitor faces the whole catalogue alone. They do not know which product suits a colleague or a housewarming, so they browse, hesitate and often leave.
Design choices
- Questions the buyer can answer: who it is for, the occasion, the budget, what they are into. No product jargon.
- Tap, don’t type: answers are buttons, so it works in seconds on a phone.
- A hard budget limit: suggestions never exceed the amount chosen.
- A reason for every suggestion, so the buyer can trust the choice and explain it.
- An honest “no match” rather than a forced recommendation, with a hint on what to change.
What the demo shows
The full flow: four questions with progress, back navigation, three ranked suggestions with reasons, and a restart. Matching runs on visible rules over 12 fictional products.
What a real version needs
Your catalogue (and ideally stock), the questions your customers actually ask, testing before launch, and a decision on whether simple rules are enough or an AI model adds value.