AI Case Study
Airbnb increased similar property listing recommendations click-through rates by 21% with machine learning embeddings
Airbnb uses machine learning to personalise consumer search results. Results are personalised based on over hundreds signals that go well beyond explicit price and room requests to the type of property listings that you click on that suggest preferences for architectural style, decor and feel. AirBnB used a novel embedding-based solution to represent 4.5 million active listings using 800 million consumer property search click stream sessions. Click-through-rates on recommended similar property listings increased by 21% and 4.9% more guests discovered their listings as a result.
Consumer Goods And Services
Travel And Leisure
AirBnB analysed over 800 million user click-stream sessions where consumers searched and clicked on properties. Using embedding modelling 4.5 million listings were represented in thirty two dimensions. The embedding could represent subtle and often non-verbalised characteristics preferences of consumers such as unique architectural styles such as houseboats, treehouses, castles. Or the feel and decor of a property. This embedding allowed search results to be personalised in real-time along with similar property recommendations.
"Airbnb’s marketplace contains millions of diverse listings which potential guests explore through search results generated from a sophisticated Machine Learning model that uses more than hundred signals to decide how to rank a particular listing on the search page. Once a guest views a home they can continue their search by either returning to the results or by browsing the Similar Listing Carousel, where listing recommendations related to the current listing are shown. Together, Search Ranking and Similar Listings drive 99% of our booking conversions."
Airbnb states that "A/B test embedding-based solution lead to a
* 21% increase in Similar Listing carousel CTR
* 4.9% more guests discovering the listing they ended up booking in the Similar Listing carousel"
A novel embedding solution was used based on the user click-stream of property listings. They encoded property listing features, such as location, price, listing type, architecture and listing style.
"Let us assume we are given a data set of click sessions obtained from N users, where each session s=(L₁,…,Ln)∈ S is defined as an uninterrupted sequence of n listing ids that were clicked by the user. A new session is started whenever there is a time gap of more than 30 minutes between two consecutive user clicks. Given this data set, the aim is to learn a 32-dimensional real-valued representation v(Li) ∈ R³² of each unique listing Li, such that similar listings lie nearby in the embedding space."
"We learned listing embeddings for 4.5 million active listings on Airbnb using more than 800 million search clicks sessions, resulting in high quality listing representations."