Launching & Scaling Uber Eats

So going back to the Uber Eats days, you know, what were— I guess, you know, when you first joined, you know, what was your role? And maybe how did that kind of change as the business really scaled? And, you know, what were some of the maybe the challenges that you were working on then?

Yeah, well, when I started at Uber, it was March of— February or March of 2014. It was just me for a long time, you know, no engineering team, no product team, nothing for the first like 6 or 7 months. So it was basically, I was almost like an analyst, if you will. And I was coming from a place, I was basically, I was out of, I was really helping lead a public company prior. And so that might give you some indication of like my draw towards more early stage stuff, as I kind of wanted to go back and do that. And Uber was, I mean, it was like the hottest thing in the world at the time, right? And so it was not the hardest decision to move there, but started sort of as an IC, then sort of built small team to run some experiments. We experimented with 3 things. We did this like delivery service thing, which is called Uber Rush, really early days, launched in New York. It was like peer-to-peer delivery. It was just delivery though. It was like, I, you know, I need to send, I don't know, a package to someone across, you know, town, or I left my phone at my friend's house, I need someone to pick it up, or all these really niche use cases. It didn't, it didn't really You know, it's not really a consumer product. And then we experimented with this thing called Uber Fresh, which became Uber Eats, which, which, you know, it was our first foray into food. And, you know, we worked with restaurants where we basically would actually put the food in the car before you ordered it. You know, Uber was highly interested in if you could deliver a car in 5 minutes, you could deliver anything in 5 minutes, you know, or what else could you deliver in 5 minutes, which turns out it's actually pretty hard to deliver things in 5 minutes. But, you know, it was an ambitious company and ambitious, obviously, you know, executive team. And so we had like a small team of maybe like 5 or 6 of us to run these experiments. We ran a convenience store thing where we literally packed a Toyota Highlander with like 250 SKUs that we found at a 7-Eleven, and you could hail the Highlander and get like toilet paper. And I, you know, I remember driving one of those Highlanders around in Washington, D.C. during the experiment, and literally it was crickets the entire day. You're just sitting in a car waiting for someone to hail you, and then someone would, and then you'd hand them a pack of Skittles, and you'd be like, why'd you order the Skittles? What could we be doing better? And they're like, I just thought this was so weird that I ordered the car and the car came to me with all the stuff in it. You know, now there's like a startup called like Popcorn that does this, and I think there's a bunch of these. They're very hard to make work, but, um, uh, you know, it's cool if it— you know, if you can make it work. One, we got really strong signal on the restaurant side with Uber Fresh, and And so that was like food on demand. You know, there are other people putting food in cars. There was like SpoonRocket and Sprig did some of this, you know, in like early days. And so, you know, the restaurant signal really spiked. And so, you know, in early stages of businesses, I'm not a big believer that consumer businesses like gradually grow from day one. It's like if you're grinding it out at low volumes in a consumer business, you're gonna be— you're gonna have a tough life. Usually you see some kind of deviation from the mean. Now, you might be working on a product and then you release a feature that makes it more spreadable or viral or whatever term you want to use. But you want to see that step change signal and then you nurture that. And we saw that with restaurants. And so when we started expanding, that team grew to maybe 50 or 80 people. We rolled out that instant delivery product. We got excited because everybody liked to use it. LA and San Francisco, et cetera, but it sort of hit a plateau. And we asked our customers, people would actually use it for lunch a lot. I mean, like, and people were really happy with it. They would get a restaurant that they couldn't get access to, sort of like the canonical example of the services. There's this, if anybody's in LA, there's this deli called Bay Cities where they have this famous sandwich called the Godmother, which is, you know, and it's, everybody knows it, but there's only one Bay Cities. And so we would have them make like 1,500 sandwiches at like 10 in the morning and it would hold up in a car. And then if you were in downtown LA, you would have to drive 2 hours to get this sandwich and deal with a parking lot and whatever, and you wouldn't do that. But with Eats, now you can get that in like 3 minutes. And so that was pretty magical. And so we found some of those sort of like getting access to interesting restaurants was an important sort of insight in early days. But our customers kept telling us they wanted more selection, and we kept telling them they wanted faster delivery. And then they said, okay, that's cool, but we want more selection. And we eventually wised up and, you know, said, okay, fine, we need to redesign this product to be more selection and lower speed. And that's, you know, at the time DoorDash, Postmates, Deliveroo, you know, a number of people were doing a version of this. We thought we could make it a lot faster, which we did. We were the fastest. Delivery in that space. But selection is really what mattered, and that's what unlocked everything. And then the team went from, you know, in 2015 December when we launched that version of Uber Eats in Toronto, I mean, literally in the first 2 or 3 hours of launching it, we did more sales in those first few hours than we did in the last week with the previous product. And so that's what I mean about step change signal, you know, in your business. And When, when we did that, we then had to prove our unit economics, etc. You know, that took a couple months to like get everybody comfortable, and then we just stepped on the gas. And so from that point, it's probably a team of maybe 100 people to when I left, it was a little over 3,000, and that was 4 years later. And so it was, I mean, we had a 1,000-person sales and ops team, probably a few hundred person engineering team. So it very quickly went from like me and like 8 or 9 people to like thousands and thousands of people everywhere in the world. And so you just have to change as a leader very, very, very fast. You go from being like the analyst figuring out the business to leading a team of people who are tinkering and enabling them to almost like a figurehead. Like when you have 3,000 employees, You're worried more about communication strategy and like goal setting and morale than you are worried about like, you know, you know, 10 basis points of margin in São Paulo.