For the first time ever, Airbnb is taking into account host preferences in its visitor-matching algorithm.
The room-sharing startup is now tracking what hosts care about, such as how far in advance a visitor has booked her trip, weekend versus weekday requests and length of stays.
Travelers’ search results will also differ. For example, if you search for a spare room for a one-week stay in Barcelona with two guests, you won’t just see the most relevant listing for your inquiry. The top-ranked results will include hosts who’ve shown a preference for one-week bookings with no more than two guests.
Airbnb has had its fair share of missteps with hosts, especially in places like New York, which has always had a contentious rental market. And like Uber, the supply side of Airbnb’s business depends on the reliability and safety of its hosts, who are essentially free assets to Airbnb and are crucial to the company’s future growth.
It was a big shift in Airbnb’s thinking to emphasize host preferences when serving up search results for visitors. “Conceptually, it took us awhile to convince ourselves this was the right idea,” Airbnb data scientist Bar Ifrach said in an interview with Re/code.
After changing their matching algorithm to account for hosts’ preferences — measured by which reservation requests hosts accepted or declined — Airbnb saw a 3.75 percent increase in booking conversion. “On average, each match was smoother and required less [request] attempts for the visitor,” Ifrach said.
This article originally appeared on Recode.net.