Is Uber Using Machine Learning To Charge Frequent Users Higher Fares?

Uber is already employing machine learning to estimate a customer's wealth to determine how much they are willing to pay.

So ride frequency on certain routes could possibly be a variable Uber is testing in some markets. JL


Aaron Mak reports in Slate:

“Route-based pricing” uses machine learning to guess how much customers are willing to pay for a certain route. A trip from an affluent neighborhood to a luxury shopping district might be pricier. If Uber is tailoring its fares based on estimates of a user’s wealth, it wouldn’t be a stretch to take frequency of use into account. “If you’ve got a high-value user and you can get them to pay more money, that’s good business. The risk is the reputation concern. Uber is already having a lot of difficulty with that.”

Last week, the popular blog “The Rideshare Guy” published a post from contributor Will Preston, who speculated that Uber had been charging customers more if they frequently use the service. Preston identified himself as a high frequency user who regularly hails rides to and from the San Diego airport, and noticed that he was being charged 10 to 30 percent more than the local mile per minute rate. He further noted that his wife, a less frequent user, often had cheaper fares for the same trip, and that there was no sign of surge pricing inflating the cost when he checked his Uber driver app. Plus, he had been riding during non-peak hours with little traffic.

Even though this is an anecdotal case, Preston says his experience leads him to believe that Uber is in fact charging habitual users higher rates. “I don’t know how they could explain that consistently my price was higher than my wife’s without them tracking this data per user,” he told Slate.

Preston’s account would challenge what the company has said about its price system in the past. In May, Uber announced that it would start using “route-based pricing,” a method that uses machine learning software to guess how much customers are willing to pay for a certain route. If passengers request a trip from an affluent neighborhood to a luxury shopping district, for example, their fare might be pricier.

In a statement, Uber denied Preston’s claims: “We do not charge people more based on how frequently they ride with Uber.”

If Uber is tailoring its fares based on estimates of a user’s wealth, then it wouldn’t have been a stretch to think they could also take frequency of use into account as well. Yet, head of product Daniel Graf told Business Insider, “[Route-based pricing] is not personalized. This has nothing to do with the individual.” Instead, he said Uber’s AI programs only look at the route itself. If a route is in high demand and ferries people from one ritzy place to another, thus suggesting richer people are requesting that specific trip, then Uber will raise the price. According to Graf, Uber is guessing how much people are willing to pay for a certain route, rather than how much a certain person is willing to pay for the service. Such a system could in effect target people who are wealthier or frequently use the service for in-demand routes, even if that isn’t the explicit purpose. We also don’t know what factors Uber takes into account when it prices these routes.

Of course, this doesn’t explain why Preston saw his fares rise as he started using Uber more or why his wife had consistently lower quoted fares than he did. Whatever the case may be, Preston writes that he eventually found a workaround for the price hikes. By changing his destination mid-ride, the app reverted back to charging him solely on time and distance rates.

Uber has long struggled with setting prices that are low enough to entice people to use the service while still making the business sustainable. Initially, Uber neglected to make a profit and undercharged riders in order to grow its customer base. Co-founder and former CEO Travis Kalanick had said these low fares were temporary, and that the company would raise them after it had dominated the market in major cities. Yet, nine years after its founding, Uber still isn’t profitable and most customers continue to pay unsustainably low fares. There’s an ever-present danger that if Uber gets rid of its ride subsidies, then it’ll lose a huge swath of customers.

Charging more for certain trips is likely a safer and more gradual way to start raising prices, though targeting frequent users would be a risky move, particularly if Uber wasn’t disclosing the practice. “If you’ve got a high-value user and you can get them to pay more money, that’s good for your business,” said Brent Goldfarb, an associate professor of management and entrepreneurship at the University of Maryland. “But the risk is the reputation concerns. Uber is already having a lot of difficulty anyway with that.”