From filters to intent

Filters tell you a budget, not a goal. The real question is which neighbourhood someone like you belongs in.

Client: Booking.com
My role: Lead Product Designer Personalisation
Period: 2024

Neighbourhood discovery.

Details from the design, at reading size.

Explore the neighbourhood
Photography and a short description introduce the character of each area.

From area to accommodation
Hotel details bring location, room information and price into one listing.

The problem I was handed

Drive personalisation across app and web. Booking's recommendation surfaces ran on filters: users told the system their constraints, and the system echoed them back.

The problem that was actually there

Filters describe constraints, not goals. Someone filtering on "Amsterdam, 2 adults, under €150" is really asking "where should someone like me stay?". That question gets answered at neighbourhood level first, and there was no surface for it.

What I did

Designed the neighbourhood recommendation feature, moving the logic from filters to intent, shipped as a carousel and a city page. Along the way I initiated the first AI integration into Booking's personalisation design work, back when an LLM in production design was not yet normal. I led journey mapping across touchpoints and delivered modular design patterns that other departments picked up.

The result

The neighbourhood recommendations shipped as a carousel and a city page. The modular patterns were picked up by other departments.

What this shows

Working inside one of the most rigorous experimentation cultures in tech teaches a discipline I still use: a personalisation idea only counts if it can be written as a testable hypothesis.