What happened

The Wall Street Journal reported that online travel site Orbitz.com was experimenting with a personalisation feature that altered the order of hotel search results based on the customer’s operating system. 

According to the Journal, people using Apple Mac computers spent, on average, USD 20 to USD 30 more per night on hotels than those using Windows PCs, and were 40 percent more likely to book a four- or five-star property. Orbitz used this data to "nudge" Mac users toward higher-priced, upscale hotels by prioritising them at the top of searches on its own platform, as well as on affiliated sites like CheapTickets and ebookers.

While Orbitz executives insisted they were not charging different prices for the exact same room, they confirmed they were showing different listings to different users by default. Since 90 percent of customers book a hotel from the first page of results, this algorithmic filtering made it necessary for Mac users to click through more options or manually sort by price to find the cheaper deals that PC users were shown initially.

The result was that Mac users were unknowingly guided to spend more money, potentially causing them to overpay by steering them away from lower-cost options they might have preferred. Orbitz's lack of transparency about this profiling undermined the expectation of a neutral, comprehensive search experience on a retail platform.

The discovery of Orbitz's opaque practice sparked public backlash, accusations of digital redlining and unfair consumer manipulation, and regulatory scrutiny. 

Why it happened

The incident was a direct result of Orbitz attempting to maximise revenue by leveraging predictive analytics and demographic data derived from user activity.

Supported by external market research (e.g., Forrester data showing Mac owners generally having higher household incomes), Orbitz’s internal data confirmed an "intuition" that Mac users were a more affluent demographic willing to pay a premium. 

The practice was a calculated business decision to increase the value of each booking based on the predicted willingness-to-pay of the customer.

Orbitz implemented its algorithmic steering system without notifying users that their operating system was being used as a factor to filter results, leading to an environment where the consumer could not make an informed choice.