Uber’s future may rely on predicting what you’re willing to pay

Uber drivers have been complaining that the gap between the fare a rider pays and what the driver receives is getting wider. After months of unsatisfying answers, Uber Technologies is providing an explanation: It’s charging some passengers more because it needs the extra cash.

The change stems from a feature introduced last year called upfront pricing. By guaranteeing a certain fare before customers book, Uber said, it provides more transparency. But it continued paying drivers using the old model, a combination of mileage, time and multipliers based on geographic demand. The difference between those two calculations could be the future of Uber’s business.

Daniel Graf , Uber’s head of product, said upfront pricing can’t be summed up in a simple formula. Uber applies machine-learning techniques to estimate how much groups of customers are willing to shell out for a ride, calculating riders’ propensity for paying a higher price for a particular route at a certain time of day, he said. For instance, someone traveling from a wealthy neighborhood to another tony spot might be asked to pay more than another person heading to a poorer part of town, even if demand, traffic and distance are the same. Uber calls this “route-based pricing.”

On Friday, Uber will begin communicating the changes to drivers. It will start reporting the price a passenger pays on each ride, though it will stop breaking out its percentage of the fare. It will also send an updated terms of service agreement reflecting the new fee system to drivers. Route-based pricing is now limited to 14 cities where Uber offers its carpooling service.

Graf said Uber’s pricing techniques have grown incredibly sophisticated. He oversees a team called marketplace at headquarters in San Francisco that’s staffed with economists and statisticians. Graf, a former Google and Twitter executive, sees financial engineering as a competitive advantage, one way that Uber can stay ahead of Lyft and other ride-hailing operators.

“Google search is very sim-