Personalizing Airbnb search by learning from the guest journey

Short summary

Airbnb engineered a Transformer-based sequence model to encode years of guest behavior—views, bookings, reviews, cancellations—replacing hundreds of hand-crafted ranking features. The system tackles three challenges: view-event dominance (97.8% of events), sparse booking signals versus noisy browsing, and computational tractability of very long sequences. The result is richer guest preference representations that improve search personalization.

• •Airbnb replaced hand-crafted ranking features with a Transformer sequence model encoding full guest journeys
• •Listing views dominate event sequences at 97.8%, creating computational and signal-to-noise challenges
• •Model learns richer guest preference representations for more personalized search results

Generated with AI, which can make mistakes.