How Uber works
An Inside/Out praise of Uber’s cultural values.
I left Uber last year, and I am still in awe of what I saw there. Travis and his co-founders have built a unique organization in its structure and culture — and I expect it to go after many more challenges in the future.
You certainly have heard of several scandals related to Uber over the last twelve months. My experience at Uber was different from what we read in the press today. I sincerely hope that Dara and the team he builds around him will work to solving Uber’s cultural issues. Yet, I also hope that some of Uber’s magic won’t disappear in its reform.
I would like to share with you three short stories to tell you about what I saw, experienced and learned while being part of the team launching UberEATS in London.
30 minutes, or it’s on us!
The story of the message that changed London from the most competitive market to the most successful launch of UberEATS.
Champion’s mindset: Ready is not enough
I was sitting in front of my General Manager and next to the Restaurant Operations Manager. Combined, we represented the three sides of the platform: the customers, the restaurants, and, for me, the couriers. The conference room had a wall that we could use as a white board to ideate. We had a great list of the 120 best restaurants in London, three times what Paris had achieved for their launch. We had applied every growth hack that had made Uber successful, from Give-Get referrals to in-app ads. We had checked every box in the playbook. But still, the executive team told us that our plan to launch UberEATS in London was “definitely not bold enough.”
Five days before Launch, we were back to square one. I must admit, we had well-established incumbents in London — after all, this was Deliveroo’s Global HQ, while Just Eat (LON: $JE), TakeEatEasy, Amazon Restaurants, Seamless (GrubHub) were all strong players. While customers generally recognized Uber’s strength in terms of Operational Excellence, the question remained: how does it impact customer behavior when ordering food?
We needed a strong message that would define what we stand for. A message that would clearly differentiate us from other platforms. When it comes to ordering food, we realized that we were, more often than not, hungry and ordering too late. So as consumers, we naturally give a high importance to delivery time. Furthermore, we already had 120 of the best restaurants in London on-boarded, and had tested all of them to make sure that regardless of what an “eater” orders, they would be happy. All we needed was to convey the message to the “eater,” so the GM came up with the idea to launch with the slogan, “30 minutes or it’s on us!”
Launching UberEATS with messaging highlighting that it is fast and reliable positioned us as a high-value offering within the competitive landscape. And we all agreed that guaranteeing a delivery time against the full cost of the meal was the strongest way to prove our point. We also agreed that “30 minutes” is, psychologically, a much stronger value proposition than any other number. Our objective was to simplify the decision making process for the hungry consumer about to order food online: if you want quality and speed, you should try UberEATS.
“Launch Day is all about Your Message — make good use of it. You cannot be just as good as the competition. You need to bring something new to the customer. Something special.” (Head of UK)
Let Builders Build: Finance stays out of the way
When I crunched the numbers, however, it appeared that we would burn our quarterly marketing budget within one week. According to data from similar cities, over 65% of meals get delivered after the 30 minute mark, mostly because it takes time to prepare food on-demand. In any other organization, the Finance team would have forced the team to change the message to 45 minutes, which — given the data — would make more sense.
In any other organization, another “division” would have expressed serious concerns that we were putting their billion-dollar business (and reputation for operational excellence) in jeopardy. In any other organization, the management team would have preferred going with “best practices” that have been tested before. And that is exactly what my MBA learnings were telling me to do. But with a culture of daring, and values to “Let Builders Build,” doing what was expected was not going to fly at Uber. The “Champion’s Mindset” had already trickled down to all ranks of the company; no manager who stayed at Uber long enough would settle for average results. Our General Manager simply took the big pile of cash we had, and struck the match.
“We will burn our quarterly marketing budget every week.”
Be an Owner: Decision Making (and accountability) stay local
Decision making is reversed at Uber: local teams ultimately decide how they launch and operate in a new city. Reversing the normal top-down approach (or HQ-Regions-Subsidiaries hierarchy) seemed necessary at first to build a company that could adapt to various environments, from NYC to Nairobi, and from Shanghai to Sao Paulo.
Moving Decision Making to the Cities also enables experimentation, as different teams approach the same problem differently. The learnings from these experiments are then added to the playbooks.
Most importantly, reversed decision making empowers local teams to move fast, as office politics and considerations of influence are limited to weekly conversations, when the General Manager has to report back.
“Speed requires local decision making and empowerment, and Uber gives City Ops teams all the resources to succeed: ready cash, motivated talent, and flexible technology.”
Make Big, Bold Bets
Burning our marketing budget in a couple of weeks was one of the best bad decisions we ever made. We instantly displaced the #2 (“Take Eat Easy”), which went out of business, after just a few weeks. Word of mouth was extremely efficient in getting Londoners to download the UberEATS app and try us. Growth curves weren’t shaped like a hockey stick, they simply looked like an insurmountable vertical wall.
From a business perspective, we decreased the average Customer Acquisition Cost by burning the cash faster because we acquired more users in two weeks than we would have acquired burning the same amount of cash slowly in 6 months. We had perfectly positioned UberEATS for years to come: UberEATS gets you the best food faster — and, they are ready to bet their cash on it.
Now we were able to compete.
“Surprisingly, we decreased the average CAC by burning the cash faster.”
Empowered Local Teams = Speed*(Creativity + Accountability)
Giving a small team of hungry managers total responsibility for a major project and the leeway to try new things, solved the classic tension between creativity and accountability — introduced by Ben Horowitz in his book The Hard Thing about Hard Things. We were creative because we had the right to fail (once); and we were responsible because the whole company was giving us everything we needed to “Make Magic.”
So we had a few more weeks before Finance started banging at our door.
In the meantime, I had to deal with a hectic start: hundreds of angry chefs with cold food waiting on restaurant counters, thousands of couriers racing through the city with back-to-back requests, and dozens of thousands of customers who didn’t receive their food in 30 minutes. Support representatives were overwhelmed with the thousands of daily complaints, while the Restaurant Ops team was getting anxious. At this stage, everyone is patient, but everyone is staring at me: will he pull it off?
“The number of meals prepared by restaurants and undelivered due to lack of couriers was increasing by the hundreds every day.”
500 meals undelivered, every Friday…
Unravelling the mess we created. A story about teamwork.
From the Courier Ops perspective, the platform’s viral success was a nightmare. Demand was four times our most bullish case. Every day I asked myself, “Where will I find the thousands of couriers needed to deliver all these orders?”
Always Be Hustlin’
It’s Friday afternoon. I’ve probably slept 4–6 hours every night since Monday. I feel both excited (#superpumped) and tired, and just survived three atrocious hours as I had to:
- Record and send an automated voice message to all couriers offline at 10.45am to let them know that we will have “an exciting day ahead” and that I hope they will get online
- Reduce the distance between restaurants and customers, and between couriers and restaurants (thereby increasing dispatch efficiency system-wide)
- Send automated text messages to all couriers who are offline to let them know about a promotion if they get online…
- …and build the promotion in the system
- Top that promotion with a general promotion for couriers that would deliver a certain number of meals over lunchtime
- Hire anyone in the office who would be willing to get on a bike and deliver a few meals over lunchtime (best way to get feedback, btw)
- Kick-off batching for meals departing from the same restaurant and heading in the same direction (basically applying uberPOOL’s algorithm to deliveries)
- Adjust dispatch parameters to send request for couriers only after the food is ready (instead of normally having the courier arrive about 30 seconds before the food is ready for pickup)
- Ask the Restaurant Ops team to close all restaurants that already had over 50 orders for the day
The Hustle.
Yet, we still had 500 meals that were prepared by the restaurant, that waited over 20 minutes on a counter, and that remained undelivered…
Now that I had used every tweak I could think of, I had to ask myself again: what’s left to solve this? I was sitting in front of my screen, baffled. My meeting with my GM was scheduled for 4pm — it’s now 3.15pm. Where do I go from here?
I had hired and trained a team of 20 partner support representatives, but they could only on-board 100 new couriers every day. I would need to hire 450 new reps in order to solve the bottleneck at the partner center, and even if I did, I would face so many other challenges elsewhere (lead time for equipment, courier on-boarding and training, etc.)
My phone was ringing. Someone calling from Paris. Allô?
“The flip side of local decision making and empowerment is Global Data Access: Everybody watches.”
Meritocracy, Toe-Stepping and Principled Confrontation
“Allô bonjour! C’est JM, de Paris Ops! You guys had a phenomenal start! Well done! I was just looking at your data, and I guess that you are stretched in terms of courier supply. I mean, you are bleeding! You have quite a few bottlenecks on your acquisition funnel. Did you consider running all background checks in parallel? Your provider must be doing them in series, it cannot take 5 weeks to process a background check. Also, it looks like you are running short with inventories, how long is the lead time with your suppliers? I can send a few backpacks in the Eurostar tomorrow if you’d like. We are lucky we won’t have any issues with the customs at the border… I mean, for now, haha!”
There it was… my solution: hundreds of other teams went through the same operational and logistical issues before me across the globe. And they were watching what we were doing, like angels. I didn’t even need to ask.
The internal systems are designed to help each city learn from others, and help others. The concepts of Toe-stepping and Principled Confrontation were introduced by Travis in order to push people to speak up and discuss matters that concern Uber’s operations everywhere. These values tie very well with “Being an Owner, not a Renter”.
As an Ops Manager, having access to the historical data of other cities was essential to our success. I could calculate the speed at which scooters move through Paris at 12.30pm, and whether they move faster than bicycles. I could review the scripts that the transportation team in Bangkok built to flag London couriers who used fake GPS location to get more dispatch orders. I could copy Toronto’s effective drip messaging campaign to reactivate dormant couriers. And we would all share a chatroom to discuss our issues and help each other over the phone.
Calls would be short, efficient, and people took strong positions when they knew what they were talking about; they would tell you what they would do and what risks they would take if they were in your shoes.
I felt there was so much to learn, and so few hours in a day.
“Uber’s IT architecture is built to promote communication between teams. Uber’s values are defined to promote communication within teams. The combination of both Makes Magic.”
Wall Street Trading Desks meet World of Warcraft
Uber’s internal systems are built like a video game. Hundreds of teams across the globe compete within their markets, and friendly-against each other. When I joined the team, our objective was to do better than Paris. Then as soon as we realized that the market was bigger than we expected, our eyes turned to New York. And as we kept growing at a tremendous pace, we passed NYC, which had just been outpaced by Los Angeles. So we had to beat LA! We expanded to London’s suburbs earlier than planned just to get to the top of the charts.
The organization of Uber therefore promotes, through the architecture of its information systems and its values-centered culture of teamwork, a wonderful competitive-gaming environment in which city teams strive to gain market share to climb to the top of the global charts, but also an environment where Ops feel part of the same global adventure.
Every day, we would look at the real-time numbers and charts comparing growth and quality metrics across teams; and every day there would be something new: Melbourne introduced cross dispatch and had thousands of additional delivery drivers on the road overnight, Singapore introduced batching and came back to profitability, Paris introduced a plan to fight fraud and increased completion rated to 99%, London expanded to 3 additional districts and outstripped New York in revenues, Austin launched today. It was thrilling.
“In Ops, we do not consider ourselves employees — definitely responsible owners and managers — and certainly also gamers — it does feel special to be part of a winning team.”
The Gang of Gamers: What’s stronger than Teamwork?
The Ops teams at Uber represent both the core and a subset of Uber’s culture: Ops tend to create their own culture, and it is followed and replicated by other teams. Ops who launched new cities, in particular, have a lot of respect for each other, and call themselves OG’s (short for Original Gangsters), given how difficult it was to launch Uber in cities where the mob controlled the black cab business, such as NYC.
This subculture within the company drives a unique sense of belonging to a group that the rest of the organization, from Product Managers in SF to Support Representatives in Manilla, could not fully apprehend. Within “Uber Technologies, Inc.,” the Ops team was the only one both in the technology world — dealing with product bugs, process automation and data analytics — and the real world — hiring drivers, negotiating with restaurants and dealing with driver issues. Travis probably participated in making the Ops team the spinal cord of the organization. This is what made Uber different from all other Silicon Valley startups.
I never would have imagined that having such a network of colleagues could help me tackle and solve the hardest problems in business. Even while working at McKinsey or the World Bank — global organizations known for sharing best practices — I never saw this level of cooperation. Within a few weeks at Uber, I had a network colleagues with whom I would speak every week to exchange ideas on making our business and processes stronger. I never waited more than four business hours for someone to get back to me. It was a chatroom full of gamers.
“The Hustlers are in charge — not the CEO, the managers or the engineers.”
Teleporters, vs. Masters of Scale
How Uber pulled a Global expansion in 3 short years. A story about technology.
For weeks, more and more couriers would congregate at the door of the most successful restaurants. They figured out, accurately, that the dispatch algorithm prioritizes the closest courier to the restaurant — just as Uber dispatch requests get to the closest driver first. This had a serious adverse effect as passing customers were discouraged to go inside the restaurant when there were 20 couriers or more waiting next to the door, and sometimes inside.
To make things worse, some couriers with Android phones started to use GPS spoofing apps in order to virtually-place themselves in the restaurant, when they were miles away, in order to get the next dispatch order. Couriers would watch fellows who just walked out of the restaurant with an order come back ten minutes later to pick up another one. No wait time for smart Android users!
The earning discrepancies were staggering. When we had just launched, the distribution of trips per hour (and, consequently, earnings per hour) followed a normal law. The Gaussian line was now slowly moving down and to the left, as most couriers were making less trips per hour, while a new skew appeared on the far right: a new group of couriers had emerged, making up to 3x what the average couriers were making.
I had been struggling with these issues for many days. I described the situation to my best friend in the office, JK, who was working on a different product and city. He tilted his head slightly to the right, raised his left eyebrow, gazed at me for a second, twitching his lips, before asking me “why don’t you place an airport FIFO zone on a square nearby?”
Technologies, Inc. — For a thousand cities
Uber had already created an algorithm to organize a First In First Out queue for drivers at the airport, and the waiting zone is a specific parking lot for all request coming from an airport area. The same function, intended for Uber’s core business, was flexible enough to be used for UberEATS couriers.
At Heathrow, drivers have to enter a specific parking lot in order to get allocated a number on the driver app. All requests from the airport are dispatched from a specific area. The same urban space organization could be applied to couriers. They would get into the queue in a nearby square, thereby leaving the restaurant’s surroundings clear. This would encourage customers to enter the restaurant, and keep couriers happy with a dispatching system that is fair, instead of random.
These ideas were still distilling in my head. I smiled, and, without saying a word, ran to my desk. JK followed and we spent the next hour deploying the airport code to UberEATS. The following morning, Singapore and New York were doing the same. The world seemed calmer. I was calmer.
“Uber’s Product team designed a system that is flexible enough to handle all situations in the real world. Ops Managers could assemble and take apart every piece of it. The same tools would be used to design processes for dozens of products, in a thousand cities.”
Hustlers innovate
There is a strong emphasis at Uber on taking action. We see ourselves as doers. We mingle with drivers. We go out of the office to talk to large groups when they are not happy with us. We roll our sleeves and do the work that we ask other team members to do first.
I however learned that, as a manager, I would be more efficient by creating process versus hustling all day to do the work myself. At our maximum level of dedication, we could only bring 16–18 hours of attention to solving problems and improving the company’s algorithm of creating value. Transforming everything I do into a process that can be taught to the rest of my team was essential for giving me the mental space to take a step back and see the bigger picture, and solve problems with higher impact.
I would typically walk out of the office a little after midnight. My late night walk home on the dim streets of London was the perfect moment to think about the phenomenal stories that happened. Some great. Some bad. Most epic. Despite the growing pain of the early days, I was always walking with a smile. After a warm shower, I would slide into my bed, place my notepad on the nightstand, switch off the light, close my eyes, and get ready to sleep. I rarely slept this fast.
Around 3 or 4 in the morning, I would extend my hand to the bed side lamp, switch on the light, write a few words on my notepad, check the time, and get back to sleep. In the morning, I would try to make sense of my scrawls, grab a coffee and run to the office. I had a new idea to implement.
“The flexibility of Uber’s internal tools moved the innovation effort from the Product teams to the Ops. Product Managers invent tools. Hustlers innovate.”
Process as a Service
We would experience new problems every day, either lunch time or dinner time, our two rush hours of the day. I would spend the day fixing whatever I can fix. I would spend the night thinking about how to solve the problem at scale. Mornings were for Google Docs.
I would start a blank page at 9.30, and would have about an hour to describe what I did the day before. I would draft a process, and ask my Team Lead to read it. We would go back and forth to make sure everything is clear, and ping every person within Uber that could have a comment on the new process: Restaurant Ops, our GM, Ops managers who worked on a similar issue, city teams that could be interested in the new process, the legal team, the strategy team, the marketing team, the support team, the community management team, you name it.
In the afternoon, I would review all comments and send to the Team Lead to get ready for training. Support teams get the training the same day, at the end of their shifts. I would make the necessary change to the system overnight, and our processes would get 1% better, every day.
“We could have hundred of team work on the same document and coordinate in real time. Processes would morph over time to make Uber’s operations stronger every day. The internal algorithm to generate value was continuously improved.”
Product Management works for Ops
In most companies, the key stakeholders are external. At Uber, the Ops Managers were the clients. If you look at the number of products that were built internally, you quickly realize that the main client is not the rider (1 or 2 apps), or the partner-driver (3 apps, including Uber Maps), but the internal Ops Manager (20+ apps). This did not happen by chance.
“The whole philosophy of Uber is to solve problems at scale, and for thousands of specific cases. Uber showed me the prowess of technology built right.”
Parting thoughts
Uber’s exceptional journey since 2009 brought something new to the world. It changed the way we move around cities. It changed the way startups enter new markets. And as the Uber mafia leaves to start new ventures, it will change how organizations work.
This story is published in The Startup, Medium’s largest entrepreneurship publication followed by 289,682+ people.
Subscribe to receive our top stories here.
