[ABNB, BKNG] Airbnb v Booking
With 60% of rooms supplied by a fragmented set of independent hotels1, the European hospitality market is set up to be aggregated: no individual hotel is large or prominent enough to effectively bargain against an aggregator or attract guests on their own, and coordinating action among a diffuse set of players is challenging. A hotel has little choice but to list on Booking lest they cede business to more compliant peers.
Many hoteliers speak of Booking as a draining addiction. But in a more charitable framing, it is an outsourced marketing channel for hotels, most of whom have no idea how to run sophisticated performance marketing campaigns on their own. As the world’s largest online travel agency (OTA), with hundreds of millions of travelers visiting its web and mobile properties every month to find, research, and book accommodations, Booking leverages scale and expertise in search engine marketing to acquire guests for hotels, who on average pay Booking ~15% of any corresponding bookings2
The same aggregation logic applies to alternative accommodations (apartments, homes, villas). Following a false start with vacation rental site villa.com in 2015, Booking has consolidated a growing number of alternative inventory under its flagship Booking.com brand. The B2B tools and sales motions used to acquire and onboard hotels were applied to multi-unit professional property managers (by contrast, Airbnb sources 90%+ of its listings from individual hosts). Hotels and alternatives were managed under the same account managers, their inventory commingled3. On Booking.com, it is hard to distinguish the alternative from hotel listings. All have the same professional veneer and all can be instantly booked.
That Booking sources alternative inventory from professional managers with multiple properties has upsides and downsides compared to sourcing from individual hosts. On the one hand, it leaves Booking more exposed to multi-homing: a single property host is far more likely to limit themselves to a single platform (the vast majority of Airbnb’s 4mn hosts only list on Airbnb) than a multi-property manager, who will work with an intermediary to populate their listings across multiple platforms. On the other hand, Booking enjoys some degree of scale economies as it is more expensive to support a hundred unsophisticated single unit hosts than one host with 100 professionally managed units.
I suspect that alternative accommodations (AA) are less profitable than hotel listings on a unit basis, given the added customer service burden that goes into managing the former (a guest is more likely to have an experience that differs from expectations at an alternative listing than at a Four Seasons). But still, the mounting popularity of Alternatives pushed Booking to aggressively fast follow Airbnb. As of June 2021, nearly a quarter of Booking’s 28mn listings came outside of hotels. In 2019, Booking generated $3.1bn in AA revenue (just over 20% of total), some distance from Airbnb’s $4.8bn but still impressive considering it wasn’t even talking about this space 5-6 years. And while Booking’s Alternatives revenue grew by just 14% compared to 32% for Airbnb, it generated “healthy” margins while Airbnb posted losses.
Sourcing large amounts of traffic through organic search used to be a point of pride for online companies like TripAdvisor. But as organic results were crowded out by paid links and Google’s own properties, this once asset became a liability and the advantage shifted to those, like Booking, who optimized around paying Google’s search toll from the start.
That Booking can convert ~$5bn of Google and brand marketing spend into ~$100bn of bookings ties back to its deep respect for experimentation and data-driven decision making4. As detailed in the Harvard Business Review case study Building a Culture of Experimentation, any Booking employee has permission to launch experiments without management approval. This is encouraged. Booking builds templates and standard reporting methods to democratize the process; trains new recruits on statistics, hypothesis formation, and experimental design; and employs a team dedicated to maintaining the infrastructure and tools that allow employees to run tests on their own5. At any time, the company is running 1k+ concurrent tests, ~25k tests per year, the real time impact on conversions, time on site, click-through-rate, email open rates, net bookings, property sign-ups or whatever KPI of interest, visible to all. The results of previous experiment can be searched by anyone in the organization at anytime. Frenetic A/B testing is now common, but I doubt any online travel company has embraced or systematized it to the same degree, as doing so requires a culture that embraces small but frequent failures (90% of experiments don’t result in any changes), and eschews hierarchy and politics.