Chapter 17, Problem 3.4ASC

Analytics at Airbnb

Airbnb is a global community marketplace that connects travelers seeking accommodations with hosts who oiler places to stay, usually at a price one sixth the rate of traditional holds. Although less than 10 years old, Airbnb has been used by more than 60 million people in 190+ countries. The rapid growth is due not only to its innovative business model of shared lodging, but also to the technology that makes the service work (see photo).

Scheduling guests into more than half a million properties posted daily is a massive undertaking. Airbnb analyzes information about its users, both the people who rent out their homes and the customers who stay there, to facilitate bookings that yield satisfied customers on both sides of the table. They then match “guests” with “hosts” (using a variation of the assignment method), display recommended sites to customers in rank order, and test the results of their match by analyzing acceptance rates and user reviews.

Further, by experimenting with multiple configurations (i.e., A/B testing) of postings, designs, promotions, and information, they can determine which ones yield belter results. The algorithms that match clients are constantly being updated, services are modified, and new variables are introduced. For example, when regression analysis showed that the quality and type of photo attached to a listing affects guest choices and satisfaction, the company offered free professional photography services to their hosts. But numbers don’t tell the whole story, so reviewer comments are “mined” to discern intent and verified when possible with subsequent slays and reviews.

Airbnb’s shared lodging model of low-cost accommodations can be copied by competitors, but the hospitality of its unique hosts is far more difficult to duplicate. Founder Brian Chesky acknowledges that the success of his business is first and foremost about the hosts. With satisfied hosts, satisfied guests will follow. Determining what factors are important to hosts is a first step to providing them with the tools to help them succeed. Data scientists at Airbnb have created a model that learns host preferences for accepting accommodation requests based on past behavior. For example, an analysis found that hosts located in larger markets tend to accept guest requests that maximize occupancy, while hosts in smaller markets like a few days between bookings. Preferences for “last minute” versus “far in advance” reservations also varied by host characteristics. These and other findings from the model affected the order in which available listings were presented to guests. As a result, both the booking conversion rate and satisfaction with the booking process saw a significant increase.

Airbnb/Stock/Shutterstock

One mark of Airbnb success? Hotels that once eschewed the Airbnb experience are now posting unused rooms on Airbnb to find new customers.

What advanced forms of scheduling presented in this chapter would be useful for Airbnb?