Uber lets customers summon a ride with the press of a smartphone button. The company has contracts with independent drivers and equips each with an iPhone and software that calculates fares and the best routes. Kalanick speaks with Brad Stone about Uber’s growth and how data trumps instinct.
How did you come to start a car-service company?
I live in San Francisco. It’s hard to get a ride. I wanted to push a button and get a convenient and classy ride.
What was the moment you realized the Uber concept was working?
It may have been when we got a cease-and-desist letter in San Francisco three months after we launched. It’s two years later, and we are still operating. Cities in California regulate taxis, and the state regulates limos. So the city has no jurisdiction over us. At the state level, there is a discussion over whether we are Expedia or American Airlines [whether they simply book cars or actually operate them]. We don’t own any cars. We don’t employ any drivers. It was clear to us that we are a booking mechanism. But we are still in those conversations with the state.
Is this fundamentally disruptive to existing taxi and limo companies?
Yes. Technology changes the ballgame.
Did you have a particular interest in transportation growing up?
I’m from Los Angeles. Probably years of my life I spent in traffic. I’m an engineer. I used to analyze traffic patterns.
How often do you yourself use the service?
Almost every day. I almost always take it home.
Give me a state of the company. How are you doing?
We’re in 14 cities right now. In July, we brought Atlanta online—a soft launch. We’ve had 26 percent month over month growth over the past 12 months—that’s in revenue. From Day One we were making money. Drivers who connect to our system are doing hundreds of thousands of driver-hours per week.
What kind of science or math happens behind the scenes when someone orders an Uber car?
We have a team here that I call the math department. We have a computational neuroscientist out of Berkeley, a nuclear physicist out of Michigan, and a couple other folks. Their mission is to keep pickup times low and utilization high. They work on demand prediction, supply matching, supply positioning, dispatch algorithms, dynamic pricing, ETA prediction.
You’ve experimented with dynamic pricing—charging more during peak demands on nights like New Year’s Eve [some customers got bills for hundreds of dollars for short rides]. And you’ve gotten negative reaction. What did you learn?
What we learned? When people get really drunk on New Year’s Eve, they need really good messaging. We told everybody, “Here’s what the pricing is going to be, push OK if you are still OK with it.” But even very small details—the exact wording—matter a lot. We need to get as crisp as possible because these are people who are not sober.
So you have not retreated from charging more during the busiest times?
We are doing it every weekend in most cities. Our principles are clear. Uber is always a reliable ride. Always. No. 2, we only implement dynamic pricing, or surge pricing, if it will increase the number of rides that happen. When prices go up, more drivers come out. When more drivers come out, more rides happen. That means less people are stranded, and more people have an option.
Some taxi drivers would say that knowing where the best streets are to pick up fares comes from intuition and experience, not looking at a smartphone.
I’ll put my scientists and our computers against a lone driver anytime. There are really good drivers out there, but we see everything. We have a lot of good data. We can help good drivers make even better decisions about how to make more dollars per hour.