How Airbnb Personalizes Search for Users It Cannot Identify

Airbnb has explained how it personalizes what first-time and logged-out visitors see. For hosts, it is a small window into how guests discover listings before they have an account.

Many visitors arrive with no login, search history or past bookings, often from paid ads or organic search. Normal personalization needs that history. Privacy rules such as GDPR and browser limits on cookies make it harder still.

Airbnb's answer is what it calls Proximity Features. It groups roughly 1,000 geographically close users into clusters based on IP location. The system then looks at what those clusters search for, including destinations, room types, price ranges and booking patterns. No individual is tracked. Dense cities get finer tiles, rural areas get larger ones, and the groupings refresh daily. Airbnb says the approach has been stable since 2023.

The results so far are modest and mostly qualitative. On marketing landing pages, personalized recommendations replaced static listing cards for new users. In the homepage search suggestions, a new user in Beijing now sees Hong Kong and Tokyo instead of Paris and Barcelona. A pilot for engagement emails is under way. The work was accepted as a paper at the TSMO Workshop at KDD 2026. Airbnb did not publish booking uplift figures.

What it means for operators: what a guest sees first may depend on where they are browsing from. Listings in destinations popular with nearby travelers could get more early exposure to anonymous visitors, so local demand patterns matter more than before.