Visualizing scale: A day of Uber rides in London on April 9, 2016
W
ith its decentralized operations teams and the shift to microservices, plus billions of dollars in fresh cash from venture capital investors, Uber was able to step on the gas in 2015 and 2016. The company had provided 140 million rides in 2014. By the end of 2015, it would complete 1 billion rides. And just six months after that, in June 2016, it would complete 2 billion rides.N And it wasn’t just providing more rides in the same cities. Over the course of 2015 and 2016, Uber nearly doubled the number of cities it served, from about 250 to 500.N
Moreover, the company was expanding into new lines of business. It would incubate ideas, then hive off promising new products, like UberEats and autonomous vehicles, and let them run fast in their own silos. Headcount soared, from 550 at the beginning of 2014 to about 12,000 at the end of 2016. Managers who came into the company supervising small teams quickly found themselves leading armies.
Senior managers in Uber’s engineering organization recounted some of the changes as the company rapidly expanded.
The three generations of Uber employees
“When I joined the company … in my group, there were five engineers,” recalled Srinivasan, who started at Uber in 2014. “Now, I lead a team of 600 to 700 people.”
Srinivasan was no stranger to rapid scaling; before Uber, he worked at LinkedIn. There, he grew a team from three people to 120 in three years.
“I thought that was pretty interesting growth in itself. But the scale at Uber is just much more than what I’ve seen than any other place. I actually don’t think Google or Facebook – which are the companies which I thought grew really fast -- still saw [as much] growth in relative business terms as Uber, in terms of going to so many different countries across the world with intense competition in every place. In the markets, Uber’s product looks the same. But it’s extremely fragmented because the political and regulatory environments in every country are very different.”
--Ganesh Srinivasan
Ryan Sokol, who like Srinivasan joined the company in 2014, considered himself part of the “second generation” of Uber employees. He sensed differences between his cohort and the company’s early founders, and with the employees who came after him — in terms of motivation, compensation, experience and expectations.
“The first generation is the Travis [Kalanick] generation — a bunch of people that stumbled upon a pretty kickass idea, got in way deep over their head, and just worked their butts off to make it happen and work, work, work. I mean at that time they're working 18- or 20-hour days, they're sleeping for four, and they're coming back the next day. I’ve never seen people work harder,” he recalled.
Those who came in the second wave, he said, “were probably more trained professionals. They were like, ‘Oh, this looks like a cool opportunity where I can exert some of my learnings and really see it in a product-market fit where I don't have to screw around with start-ups. I can really exert my energy.’”
Sokol had been at start-ups before and had a sense of what to expect. “I came here willing to put in my four years and grind myself to paste if I had to, because it was fun, it was engaging, and it was exciting,” he said. “Most of the people that came in the second generation of Uber came for that same reason.”
The next wave of employees, said Sokol, was motivated as well, but some were surprised by the pace and the workload. “When they got in here and they saw that it was charging so hard, I think some of them started to think, ‘Oh man, what did I sign up for?’ ”
Uber had a performance review system that involved “stack rankings.”N This practice was popularized by Jack Welch at GE in the 1980s and was used by many large companies for years. With stack rankings, only a small percentage of employees, typically about 10 percent, could be designated as top performers. Meanwhile, a set number had to be labeled as low performers and could be fired or pushed out. Some people referred to this system as “rank and yank.”N
Stack rankings started to lose popularity around 2010. According to a survey of large companies by the Institute for Corporate Productivity, 49 percent used stack rankings in 2009, but by 2011, only 14 percent used them. Microsoft gave up the practice in 2013.N
Rankings at Uber were given twice a year, and performance review scores were used for compensation and promotion decisions. N
Uber would later face several lawsuits over its pay and promotion practices. N
As Uber grew, Pham’s engineering division promulgated five main expectations for its managers: vision, leadership, team building, judgment, and execution. Srinivasan explained:
First, we look for a vision. Do you have a clear idea of what you are trying to solve, why you’re trying to solve it? Second, leadership — the ability to inspire or motivate people. Third, team building. You have a vision, you’re a strong leader, but if you don’t have a team you can’t do anything. How do you build a team? How do you coach them? How do make sure they’re a really good team? Fourth was sort of a unique Uber thing, we called it truth-seeking. The idea was we were growing so fast, the best ideas can come from anywhere. Let’s make sure that we are spending our time exploring ideas and let the best ideas win. “Ideas over hierarchy” kind of stuff. And then the last one, which was last but not least, was deliver results. You have to solve problems for the company and actually deliver. The last one is what you delivered and the first four are how you delivering it.
Around the same time, Kalanick and the senior leadership rolled out 14 values for the company as a whole. These were not shared externally, but were widely known internally.
Uber's Values (2015)N
Make Magic
Click here to watch a video of Travis Kalanick talking about Uber's values
Some of these were conventional, like “Customer Obsession.” But others were very particular to Uber, including one known as “Let Builders Build.” A number of them overlapped with the engineering division’s five core competencies, while others did not. As Uber grew, some of these values served the company well, others less so. Pham recalled how “Let Builders Build” went from motivational to detrimental over time: