In just four years, Airbnb had become a premier online marketplace where home owners and would-be bed-and-breakfast owners could turn vacant rooms into nightly rentals. Potential guests could simply log on and search for available rooms, transforming any house into a possible hotel. By July 2011, Airbnb was valued at $1 billion and had rented over 2 billion rooms since operations began in 2007.
Airbnb created an online reservation system to facilitate transactions, which included a reputation system to build trust and reliability among market participants. In particular, a longstanding review system let renters read about hosts before submitting a reservation request. Similarly, hosts could read about renters before accepting.
Most of the resulting reservations proceeded without incident, and in 2011, Airbnb was seeking roughly $100 million in a second round of venture capital.1 But in June 2011, a surge of criticism called into question Airbnb’s entire business model. A corporate events planner going by the pseudonym “EJ” had rented her San Francisco apartment to a guest booking via Airbnb. They had never met, but followed the Airbnb protocol and had exchanged several messages through the Airbnb system. The guest seemed perfectly friendly, and during his rental stay, he even sent EJ messages thanking her and complimenting the “little loft area” in her apartment. When EJ returned, she found her home ransacked and her belongings stolen, with damage totaling tens of thousands of dollars. EJ began blogging about the incident. An early blog entry described some of the irreparable damage.
They smashed a hole through a locked closet door, and found the passport, cash, credit card and grandmother’s jewelry I had hidden inside. They took
my camera, my iPod, an old laptop, and my external backup drive filled with photos, journals . . . my entire life. They found my birth certificate and social security card, which I believe they photocopied—using the printer/copier I kindly left out for my guests’ use. They rifled through all my drawers, wore my shoes and clothes, and left my clothing crumpled up in a pile of wet, mildewing towels on the closet floor.2
EJ’s story spread quickly, accelerated by the attention already resulting from Airbnb’s rapid rise. Negative press was featured in the Wall Street Journal, New York Times, San Francisco Chronicle, USA Today, and ABC News, among other media outlets. The situation was made worse when Airbnb was slow to respond. Meanwhile, a second victim came forward saying that his apartment had also been ________________________________________________________________________________________________________________
Professors Benjamin Edelman and Michael Luca prepared this case. This case was developed from published sources. HBS cases are developed solely as the basis for class discussion. Cases are not intended to serve as endorsements, sources of primary data, or illustrations of effective or ineffective management.
By July 2011, the uproar had subsided, and Airbnb proceeded with raising $112 million of venture funding from Andreessen Horowitz, DST Global, and General Catalyst Partners.4 But users were rattled by the experiences of EJ and others, and some critics questioned whether concerns about trust could undermine Airbnb’s model. It was clear that Airbnb’s online review system was not sufficient to protect users and preserve confidence. Airbnb managers had to do more to regain the trust of market participants—especially hosts. But how?
Founded in 2007, Airbnb created a forum where hosts could post room listings, along with descriptions and photographs. Interested guests could search for rooms in a given area and examine possible options (including description, photograph, and reviews). Guests could filter results by price range, neighborhood, availability of a room versus an entire apartment or house, and other characteristics. (Exhibit 1 shows a listing for an available room. Exhibit 2 shows examples of reviews.) Since reviews were the primary mechanism for verifying the characteristics of a room, Airbnb placed reviews prominently on each room listing.
After choosing a preferred room, a guest submitted a reservation request to the corresponding host. The host then had the opportunity to accept or decline the guest’s request. Before accepting a reservation, a host could see the profile of the guest. Some guests uploaded pictures to their profile, and guests who had used Airbnb previously were noted as such (with a listing of their prior reservations). A host was asked to respond to a request within 24 hours. If a host did not reply to a request within that time, the request would expire. Further, Airbnb included on each host’s room listing the proportion of reservation requests the host had responded to within 24 hours. A host’s placement in search results was also affected by the proportion of requests that the host had accepted. This provided the incentive to respond to requests in a timely manner. Before accepting a request, the guest and host could exchange messages through Airbnb but were not permitted to exchange e-mails or phone calls outside of the system until after a reservation had been made.
To make sure that all communication prior to a reservation occurred within the Airbnb system, Airbnb developed an algorithm to block messages that contained phone numbers or e-mail addresses. If a host accepted a request, Airbnb authorized the payment and provided the host with the guest’s phone number and e-mail address. If a host declined, the guest was notified. Airbnb charged a fee to both guests and hosts. Guests paid 6% to 12% of the reservation cost, depending on the total cost. In addition, hosts paid a 3% fee for each booking. Guests paid Airbnb by credit card at the time of the reservation. Airbnb held the guest’s payment until the reservation began, and within 24 hours of a guest’s arrival, a host received payment from Airbnb by check, direct deposit, or PayPal.
Online Reputation Systems
By enabling low-cost communication and matching between buyers and sellers, the Internet facilitated the creation of myriad new markets. A recurring problem in many such markets was missing information—market participants lacking the information they needed, or felt they needed, to choose appropriate transaction counterparts. For example, a buyer on eBay or Craigslist often worried that the seller’s merchandise was not as described. Even when previous buyers had reviewed a seller, buyers worried that the reviews were uninformative (perhaps because reviewers had different preferences, because the seller’s practices changed, or because the reviews were fake).
Meanwhile, sellers also had concerns. For example, the consumers who decided to review a restaurant or hotel might be quite different from the usual customers of those establishments. Online marketplaces handled these concerns in different ways. From its launch in 1995, eBay included an online review system where buyers and sellers could review each other. Research showed that sellers with better reviews were able to command higher prices for a given product.5 At the same time, there were many concerns about eBay’s system. For example, sellers could open multiple accounts. As a result, unreliable sellers could simply switch to a new account after receiving a bad review. Further, the reciprocal feedback system (buyers and sellers rating each other) raised concerns of collusion among reviewers.
Dedicated review websites such as Yelp allowed for reviews of products provided entirely offline. Yelp had greatest usage in reviewing restaurants, but also other services ranging from dentists to clothing stores, supermarkets, and spas. Yelp created a social network of reviewers, allowing people to post profiles about themselves and become online friends with each other. Yelp also rewarded reviewers who were especially prolific, even offering periodic parties for “elite” reviewers. Reviewing hotels and other travel-related amenities, TripAdvisor added structure to collect information about reviewers: TripAdvisor asked reviewers to identify whether their trip was for business or leisure. Reviews for leisure travel were further broken down into family- and couplesoriented reviews. Users browsing reviews could focus on reviews from others like them.
Improving Airbnb’s Reputation System
After the two widely publicized summer 2011 incidents of untrustworthy Airbnb guests, it was clear that Airbnb needed to do more to facilitate safe transactions. To improve safety at its marketplace, Airbnb could implement the following options, in any combination.
Present More Sophisticated Consumer Reviews
While Airbnb solicited consumer reviews and featured these reviews prominently, Airbnb’s review system lacked some features familiar to consumers from Yelp and TripAdvisor. For example, Airbnb included no notion of “elite” reviewers, and Airbnb offered no reward to users who submitted reviews. In principle, adding features to the review system could help to more quickly identify problematic hosts or renters. Of course, there would be myriad decisions—whether to create tiers (and if so, how many and with what criteria), what benefits to offer, and what other classifications to include.
Collect Other Information about Hosts and Renters
Before accepting a reservation request, hosts were able to observe a customer’s profile. However, profiles were often incomplete and contained little information. Further, guests were not required to provide full names upon submitting a reservation request. Airbnb could encourage or require more complete profiles—providing hosts with pictures, names, and other information about potential guests. Hosts would then have additional information to consider when deciding whether to accept a request.
Make It Easier to Reject Renters
Airbnb had a policy of penalizing hosts who turned down reservation requests for a posted room: the more frequently a host declined a guest’s request, the lower the host appeared in the search results. Airbnb had established this policy to encourage hosts only to post rooms that they were actually willing to rent, and to help guests find the rooms that were most likely to be available. However, some hosts complained that they felt pressured to accept reservations even when they were uncomfortable with a guest’s profile.
Vary Prices Based on Ratings
To date, Airbnb had required hosts to post a single price for a given room on a given night. Airbnb could allow more sophisticated pricing to address differences between guests’ ratings. Then hosts could accept guests with little reputation or mixed reviews, but charge those guests a higher price.
Provide Insurance for Hosts
Airbnb could reimburse costs associated with unreliable hosts or damage caused by guests. EBay had implemented similar buyer protection to reimburse consumers who did not receive items they ordered on eBay.
Guests who had a good experience might become repeat customers. Airbnb could build trust by paying for a portion of each customer’s first rental. Consumers who had a bad experience were at risk of abandoning Airbnb. But Airbnb could pay a bonus or refund to consumers who reported a problematic stay.
Prescreen Hosts and Guests
As of summer 2011, Airbnb did not prescreen guests or hosts. Airbnb could add a screening process. Existing background-check services could verify a person’s name, address, phone number, and other details by confirming with public sources and even cross-checking with credit records, arrest records, and even results of Google searches. Or Airbnb could add staff to interview users and examine properties in key cities, a technique recently added by several dating websites.
The Decision at Hand
EJ’s experience had sparked complaints from other concerned hosts. A Gawker story described “Airbnb horrors,”6 and Airbnb’s detractors—already raising questions from sales tax to unlicensed hotels—were quick to highlight these problems. Airbnb needed to fix these problems and assure user confidence. But how?