The Inside Story to Uber's Hypergrowth Scaling; What Worked, What Did Not?

I was in digital marketing very early on, so almost twenty years ago. My first client ever was Sun Microsystems. We were acquiring Java developers to make apps for phones that looked like candy bars at the time, not iPhones and things like that. And really grew up in this world of very data centric marketing, where everything that we did could be measured and thought about. Inevitably, my kind of path became this emerging media path where I’d helped clients understand how do you leverage new technologies, social and mobile and video.

And for a lot of those traditional clients, so Budweiser, JC Penney, etcetera, it was not just thinking about marketing. It was also thinking about the product. How does the product need to evolve and be different? You can’t just take what you had on your website and throw it on a mobile device or put it on Facebook or things like that. And so that to me is feels a lot like what I do as growth now or a precursor of it. But Uber was the first place that we called it growth.

I was hired by Ed Baker. He brought over the kind of Facebook structure that they had built at Facebook. And so that was the first time I had a title and thought of myself as part of a growth team instead of just a marketing team or a part of the business.

Yeah. I think it’s absolutely a mix of the two. The core of growth really dives into user psychology and understanding customers, making sure that you’re actually connecting with people when the perceived value of your product or your offering is at its ultimate highest, and then building technology and building systems around that to soak it up and actually be able to manipulate it and do it over and over and over and over again. So to me, that’s the art is the people and the science is the data and machines and systems.

Yeah. So Hotel Tonight was a mobile only travel app. And the relationship relationship that that we we had had with with our our hotels hotels was was one one that the rates that we were marketing could only be done on mobile devices. It was a kind of a rule left out of a lot of the early paperwork with Expedia and the big traditional players. And that meant that we couldn’t leverage SEO. And that’s really how, like, most travel companies had grown up into that point. And so that really forced us to be at the forefront of anything mobile growth.

We were doing a lot of incentivized ads to jump up into the app store. We were doing a lot of the early versions of mobile referral programs with contact lists. It really just honed in my ability to understand and think about new channels. One of my like happy moments to look back on is that I was working with the Facebook product team and was the first customer to use mobile ads on Facebook. And so actually being at that forefront, the only reason we were able to do it is because we had a good brand and we weren’t a mobile game.

And so Facebook wanted a different example other than games in that mix. That kind of new channel ability and thinking outside of the SEO box for hotels was a really big opportunity for us.

Really thinking about and driving growth where growth really impacted both sides of the marketplace. Having to do it where the marketplace is hyperlocal and changes immediately, like every moment. The marketplace has a different dynamic. I think that just became a masterclass. The MBA in marketplaces and really understanding the dynamics of, okay, we’re not just thinking about customer acquisition. We have to think about how the relationship between the rider and driver or the restaurant and the rider or the trucker and the shipping company and really thinking about marketplaces was just a hardcore element of helping to grow Uber.

So actually similar to Uber where I started and like literally halfway through my interview, it was about how do we get more riders. Travis came in and was like, wait a minute. What about drivers? I had to shift and think about it very differently. Lambda School was similar where it was like, hey, how do we get more students into our school? And then we quickly realized that Lambda lives and breathes on getting people jobs, not getting people into a school. And so we shifted all of our product and marketing work to the the job placement side.

And that was just this unique moment where you’ve got all of these different customers using the exact same product. You have students. You have teachers getting them into that point. You have student coaches. You have salespeople, and you have the hiring managers. And so that building product and growth around just a hugely instrumental part of the company’s success and a complex moment was really enjoyable.

Apple really cut Facebook’s knees off and made hyper targeting ads really difficult suddenly, and that was really the bread and butter of how Masterclass grew for a long time. People started going outside again post COVID, and Masterclass get this inflection point where a single instructor was no longer this whole, wow. That’s crazy. You’ve got that instructor on there anymore because all of the instructors they had were that. And so we had to shift from, hey, how do we take a single instructor and market them and leverage them to get somebody in and using our product to really thinking about the value of the portfolio of Masterclass and the entire subscription.

And so Masterclass just became this hardcore element of how do we think about engagement and how do we think of the portfolio value of our product and really shifting that awareness of our product to the customer

It was pretty mixed. I would say the majority still comes from Masterclass first. I think a lot about launching a class, a masterclass similar to trying to get on the New York Times bestseller list, where you use the leverage of that talent to hit that spark right out of the gate to get the biggest audience in one fell swoop. That’s where that talent network would really come in handy. We also had the ability to do, like, white labeling and things like that, where we’re able to leverage Gordon Ramsey’s Facebook page to do marketing and that type of stuff, which was definitely an advantage that a lot of companies don’t have.

But I would say that burst at the beginning was valuable, but you still got a lot more value over the course of time from an individual class than just

Yeah. So the two very different things in my mind. I’m on a bit of a high horse right now about this idea of a growth CMO. And let’s just talk marketing for a second. When we think about marketing and building marketing teams, I find so many companies get stuck in this trap of, do I want like a performance driven CMO or do I want a brand driven CMO? And what I believe that does is infuse this idea that you actually have to think about those two things like wildly different.

But when I think about growth, whether it’s a header growth or a growth CMO or whatever, it’s more about the art of understanding how to leverage all of the data, all of the flexibility of the real world economics of our world that we live in now, and technology, and actually be able to implement that. And to me, that’s not only valuable for paid ads or SEO. It’s also valuable for how you think about the design of your business and your brand and the story you’re telling and the content you’re creating.

To me, a growth CMO is actually like a full stack CMO, but actually leverages the idea of, hey, let’s bring in data experimentation and product and engineering to help us drive the entire system, everything from your performance media to brand and PR on the other end of the spectrum.

speaking, I actually don’t think most companies need a growth team. Growth again should be infused throughout the business. So quick story in terms of Uber. Right? We was hired by Ed. We formed the growth team, and we just got funding from Google. We’re already a $2,000,000,000 company. We had 300 employees. Most of those were ops people at the city level. But so let’s say it was a 75 to a 100 person kind of centralized engineering driven organization. But we had massive product market fit. It was very obvious at that point.

We built this growth team because the core product team was knee deep in trying to figure out how to grow our product. The core product, we needed someone to be able to focus on growth. That quickly evolved into two different growth teams with rider and driver, and then an international growth team, a China growth team, and then eventually it disappeared. All of those functions within the growth team eventually actually merged back into the core business to where by the time I left, we actually didn’t have a centralized growth team anymore.

I think people assume that Uber succeeded because they had a growth team pre seed and built it out and they still have it and it’s working magic and all this kind of stuff. It’s no. Most of the functions within growth for a normal growing company can actually just be built appropriately within their associated teams within marketing, having performance marketing within product, having people owning the top of the funnel acquisition or activation, retention, engagement. So most companies actually don’t think need a standalone growth team.

never actually been on a product or a marketing team that didn’t care about growing their business even if it didn’t have the title growth. So I think this idea that there are teams and companies that don’t within companies that don’t care about growing their business is a fallacy. I think the real risk is it being deprioritized for noise or whatever the hottest thing in the market is right now or something’s broken, so we have to fix that.

So if you do build a growth team, right, I think there’s this element of having a a strike team nature to it, a full stack team with product, with marketing, with design, with data, with engineering that can go and dive deep into the core point of leverage for your business to grow right now. Maybe that’s acquisition. But once you’re done with that, it’s not like you stop doing acquisition. Maybe it’s activation, or maybe it’s a referral program, or maybe it’s whatever it could be. Anything that’s good and worth it has to keep getting worked on.

With the way you build a growth team, it’s typically not the right DNA of people to just keep working on those things. This idea that you have to have a piece of the business carved out, and that’s only thinking about growth. I think that’s where a lot of people get stuck and have trouble churning through heads of growth or being like, oh my gosh, where should this sit while the core product leader and marketing leader are like, we’re doing the work right now. Why do you think we need another person?

It feels very reactive to thinking you need growth rather than I think the best growth teams, the stand alone growth teams come from when you have hypergrowth and you need to control it and contain it and understand it. That’s when I think focused growth team is like, hey. Figure out what’s happening here and how we can productize this and keep it growing and make sure it’s not just like a flash in the pan. Okay. You’ve thrown my schedule into completely May. It still works. I think there’s still value there, but I’ll probably come back to that a couple times.

Think especially early stage companies thinking, oh my god, I need a head of growth. I think it’s a very like risky proposition. Do you think hiring heads of growth too early is the biggest mistake founders make? Yes. Yeah. I would say so. Because again, I think that founders are like, we’re not growing fast enough. I should hire a head of growth. And if you’re not growing fast enough, it’s usually because you don’t have product market fit. You actually need to improve your product first, and, yeah, still grow, but you’re looking for hacky ways to grow your business that actually isn’t working in the marketplace yet.

I think scoping it is the first stage. So let’s look at what team do you have right now, and let’s say what do we want the team to be a year from now? What do we if we grow the way we think we’re gonna grow by hiring this person, what team and functions do we think we we actually need? So one of the challenges I see a lot of people do is they’ll hire one growth person. And as I mentioned, when we built growth at Uber, we infused an entire team in one fell swoop.

Ed was hired, and he hired myself, a head of product, a head of engineering, a head of design, head of analytics. And so we each were operating our own individual functions and able to grow as a group, not just, hey, Ed now has to go and barter and trade to get resources from the rest of the organization. So I think that mapping of saying, okay, let’s say this really works and this is we’re making the investment in growth as a standalone thing. What does it look like a year from now?

And in that process, I think that’s where people really often realize, oh, what we actually need is someone to really think about our acquisition and activation funnel. And maybe that is really just a growth product person that sits within the product organization that is thinking about those elements of the team. So I think that exercise of like thinking through a year and scoping it out can help you really understand of do you really actually want this third party team operating independently of your growth and marketing teams a year from now?

And what does that even look like? And that’s where most people are like, oh, I’d actually rather this be embedded here and here. So maybe you still hire a growth person, but hire them and be like, hey, do this hacked together growth team for six months. But the goal is actually for you to land as one of our senior product managers on the product organization where you’re gonna run this. And then by the time that happens, we’re also gonna have a head of marketing and so maybe that’s where performance marketing will sit.

And again, all that could change in those six months, but those exercises to me help you actually, one, have a realistic conversation with yourself of what growth needs to do for your organization, and then scope it and have the right conversations with people when you actually go to market and start recruiting.

So probably the two most common. So one is I have an exercise that it came from Uber, which is how do you determine what you’re willing to pay for a rider or a driver in any given market, which is like a very complex marketplace. Yeah. Go ahead. Answer it. Do know

So the challenge is that the LTV is usually based on the value of a trip. But that value of a trip has a shared relationship between the rider and the driver. And so you have to decide like the revenue that you get from that trip, if you’re getting 20% revenue from a $10 trip, you’re giving that $2 calculate the lifetime value of the rider or the lifetime value of the driver? Vohra? No. The rider. The rider. The driver is doing many trips, so they have a higher LTV.

Exactly. So we could probably spend another thirty minutes going back and forth on this. But this it just becomes a very complicated quite answer because it is both. You at the ultimately have to decide when you’re marketing to both of those customers, when I have to pay to go get a rider and I have to pay to go get a driver, I have to calculate both of their LTV. So I can’t just give it to one side of the marketplace. But then you have a dynamic where, like, some markets you need more riders, some markets you need more drivers.

So you may be willing to pull more of the value over to the rider or more of the value over to the driver. It just gets really complex. And there is this is one of those challenges that, like, we’ve had we have dozens of data scientists working with it, and we never got a great answer to it. But just hearing how people think through that and challenge a push, oh, okay. I see that. I think to me, like, really highlights the ability, again, to wield data, to be able to think through the complexities of it, not just take another model that you heard somewhere else and apply it to something you’ve never most people haven’t thought about that specific problem.

So the other question is, I like to dive into engagement. Because a lot of people in growth, they think about acquisition. And so I typically give some type of product where I’m like, look, we’ve actually got really good, consistent engagement, but now we wanna grow engagement of these users. So I want you to think through examples of how could we get new customers engaged, how can we get more frequent engagement, and how can we get more quality engagement, come up with examples or how you would test each of those.

And then usually I go through a let’s think of this like a product manager. How would you prioritize those? How would you pull in the resources to test? How would you test those? And how would you pull in those resources? Again, I find that gets the double hit of how do you think about the real customer need down funnel, not just at the top of the funnel? And how do you think about the operating these types of experiments that aren’t just running ads or running landing page tests?

So the biggest one for me, and this is very much my marketing hat showing, is people that just don’t talk about the customer. And what is the actual customer problem? And how are we solving it? And what’s the value of it? When they just are like, all we’re gonna think about is the data. Or again, just pulling in, oh, here is how I did it exactly at this company, where it’s copying and pasting what you did at a previous company is to me just like a red flag of someone taught you how to do this rather than you actually having the framework in your head to go apply it to the unique nature of my business and my customer.

And so those to me are just big red flags that I see. I run into a lot with people that I find to be it could be really great talent to have on a team, but are often more suited for a very specific role that has a very actionable, you’ll get this thing done, it’s very directed to you versus a growth role, which I think is like a very, like, pioneer exploratory role.

I like to look for a couple things. So one is, what is something that can rally as much of the company as humanly possible? So you could say something like new customer is our North Star. There’s huge chunks of the business that really are whatever. We’re gonna do our job whether we get 10 x the same number of people or we slow down. The job is exactly the same in front of us. I also think that when you’re dealing with marketplaces, having a metric that actually forces you to think through which side of the market needs more growth or more effort.

So at Uber, I think you and Ed talked about this around the trips was our metric because it’s okay, a trip can’t happen without a rider and a driver. And so if trips slow down, it forces us to go into the weeds and say, why aren’t they happening? Is it pricing? Is it the amount of riders in the market? Is it the amount of drivers in the market? Is it where they’re going? Like, it just forces you to ask more thoughtful questions. And then the last thing I like to think of is actually rooting it to the customer problem.

A lot of Silicon Valley uses like DAUs or MAUs or things like that, but most people don’t actually use those products. The problem that their product solve doesn’t happen daily or weekly or monthly. It actually has a once every three weeks or two times every six months. Masterclass isn’t when you’re in a class, it’s maybe a daily active, but then you’re maybe gonna take three classes a year. You’re not gonna take another class the second. You’re done with the first class and then the third class and the fourth class.

And so actually routing it to, like, what are they coming here to solve? To me, that as a North Star metric, make sure that you’re thinking about the actual engagement and retention elements of your business appropriately as well.

This is another one that can be a trap for some people in the sense of, hey. I have to have a really, really crazy, well defined model and everything, primarily because everything changes so quickly, especially for early stage founders. That being said, what I typically like to do with my founders is create what I think of as a vanilla model. Here’s like the four step process that you’re gonna grow. You’re going to get more customers. They’re gonna create a piece of content. That content is something they’re gonna share with a friend.

That friend loves the content so much that they want to become a customer, so they share. I’ve created some four step loop and like using that then as an anchor for two major things. So one is, do you actually connect the dots? Do they come back at full circle? Do the things that your customers do on your platform give a reason for the next person to actually become a customer? If not, then like you’re actually you’re growing a linear business, which is fine, and there are lot of very successful linear businesses.

But know that you actually are going to have to probably do more paid or do more really traditional top of funnel just getting custom sales motions, things like that. The other piece is it gives you four elements that, like, every planning meeting you go into, every prioritization meeting you have, you’d be like, is this actually impacting one of those four things? And if it’s not, this probably isn’t helping our business right now. But I think just a simple loop of your business to start is a really great healthy way to anchor all of the efforts that you’re doing and to make sure you actually know whether or not your business has an a k factor or not.

Easier to run is the concentrated. I find it incredibly more fun to run the distributed long tail supply. That’s my one short answer.

thing. It was dozens and dozens of things. Just a couple quick ones. So one was a new channel discovery. The idea of onboarding and acquiring hundreds of thousands, millions of drivers is something no one’s ever done before. There’s no case study. Literally, like, every other week, we’d be like, okay. We were just spending time with our ops team, and they post stuff on Craigslist to get drivers to show up in our office to sign up. Could we automate that? Could we actually take that into our growth team and make it a scalable channel?

Oh, Craigslist works. Wouldn’t all job boards work? We had to go into Indeed and Monster and actually, they’re used to posting two jobs at a time. We’re like, no, want hundreds of jobs for every category. We want to treat it like Google, not like a job classified. So just only iterating and thinking through, hey, we actually have to just be the snowplow here and go and find and create and identify ways to get drivers on board. I think the second thing was just really understanding, taking the time to do just tons of customer research.

Driver is such a broad array of people. You have people that are professional drivers already, taxi drivers, things like that just shifted their work effort over to Uber. You had people that were like, hey, I’m doing odds and ends jobs as it is. I’ll just add this to my mix. You have people that weren’t working. They’re maybe stay at home moms or had schedule that they just are like, oh man, having one more thing to plug the hole in. You have retirees that are like, I just want to talk to people.

I want to hang out with people. And really understanding those because initially we were able to just go market the heck out of, hey, you can make a thousand dollars a week. Okay. Everybody wants that. But eventually, that’s not enough to build trust for someone to be comfortable with someone in the back of their car or for for that retiree. A thousand dollars a month isn’t that interesting. It’s actually the communicating, the engaging with people, getting to know your city, things like that. And so having this just we brought on customer research onto our growth team really early on, and it was how do we unlock that new fragmented piece of our marketplace?

Oh, I’m

Which one is the most memorable one? You frowned. They So when you would interview drivers who’ve been on the platform for a while, the thing that they would always say was that they don’t actually drive for the money, they drive for the flexibility. Flexibility, when Stack ranked, money was like number three. The flexibility, the ability to just literally turn on my phone and I can go make money right now. I’ve never had anything that easy, that accessible. And so we took that insight and just went we need to, like, shift all of our messaging away from dollars to flexibility.

And we spent a ton of money on that type of messaging, and then it just burned. It was just like no one clicked. Right? And so what we learned is that the messaging the point of time for the messaging really mattered. The money was the thing that got people to be, oh, that’s interesting. I’ll go click on that. Whereas the flexibility became more of an engagement strategy. And so taking an insight from your active customer and applying it to new customers was just a big miss for us on, like, the marketing and messaging side of things.

that

Yeah. I think you should. I think you should just you have to be really thoughtful about who you’re researching and who you’re researching and where you wanna apply that research. If I’m researching researching and I’m interviewing people that are active customers, and I’m gonna go apply that research to inactive customers, I can’t assume that the same insight is going to stand and

Both. So I typically try to build some type of pyramid messaging structure that on the bottom of the pyramid, the broad stuff is you can make money. Okay. That might still apply to that retiree. That might still apply to both men and women. And that I can put into mass mediums. I can put that on my my on uber.com. I could put that in TV ads, things like that. Then the next tier up might be something more specific. So let’s take the safety and targeting female audiences.

And so it’s okay. We know that women, another very large group, but gets a bit more specific. Safety really resonates there. So now I may take those channels and put that message in I may still be on TV, but now I may, not target sports. I may negative target things that I know are gonna be very male skewed or I’m gonna put the right talent in there that actually speaks to my customer. I’m gonna put that on it may not be on ubercom, but maybe there’s a new badge on uber.com that says learn about safety, and, like, it goes down that path.

Then it may be, what about safety in Chicago? Or it’s, oh, we’re seeing that safety matters more in Chicago than any other market. I lived in Chicago for a long time, so I’m picking it not because I think of it as a inherently dangerous city. But let’s just say that’s a thing. So it’s again, you may build content around that, but now you’re gonna find channels that are very specific to Chicago and that message hitting there. Ideally, you wanna build a structure that allows you to put the right message at the right place.

And the better that targeting can be, the more refined the messaging can be. Ultimately, you just have to figure out, okay, safety is a pillar that we can tackle. But safety for retirees in Boca Raton is probably there’s just not enough market there for us to build an entire creative flow for that.

data matters. So at Uber, we can learn pretty quickly. We were spending a billion dollars a year at one point. For me to test something in the ads layer, I could get learnings within minutes. And do you pull it then in minutes if it’s all nothing new? Potentially. Most of the time we were, again, building like creative systems. It was just like, hey, we’re gonna actually test 50 different messages now within this creative system within a week, Rather than just testing one thing for a week, then having to test the next thing, we had the volume that we could test a ton of different things at once.

Rather than putting all those impressions against one message, we could separate it and actually get a lot more insight and learnings into the direction we were going. Whereas like Lambda School, if you think of the funnel for Lambda School starts with acquiring a student, then they go through an admissions process, then they’re a student for nine to twelve months, then they you’re interviewing for a couple months, then they get a job. So it’s literally a year long funnel before you know if something really genuinely worked.

And the volumes we were talking about are like a 100 people in a class, a couple 100 people in a class. Right? But we were spending maybe like a $100 a month and not just tons and tons of money. And so that that cycle became, okay, let’s try to find signals. Let’s pull in a lot more qualitative research. Let’s talk to the students, see how do they reflect with that messaging. Do they think it’s gonna work? And that just naturally would be more that environment where we’re gonna test one thing for the next three weeks, see if we can get enough signal in it to keep investing in it.

So it can range pretty drastically.

cohorting is really important when thinking about retention. I think a lot of times people get stuck in just looking at our monthly actives going up and things like that, and really looking at different cohorts either on a weekly or monthly basis. And then again, going back to what is the problem that you are solving, and what is the natural use case of that problem. And so at Uber, if I took a segment that I knew were commuters, where I’m like, I would expect this cohort in this segment to use our product three times a week.

Maybe not every single trip, but they’re gonna use it three times a week. We’ll pull in enough data to do that analysis. We’ll do some, like, histograms to understand what is the normal average, and then using that against that cohort to see, okay, does that three trips a week, is it happening regularly month over month or week over week? And then that gives you both, okay, this is a retained user, and it gives you the anchor to say, oh, if they drop down to two or drop down to one a week, then that’s someone I could actually start engaging with.

You you don’t wanna have to recapture a user that’s totally churned. You wanna see someone, okay, they’re looking like they’re gonna churn, so what’s our strategy for those people? Or are they now using it 10 times a week? Okay, they’re now in a totally different segment, so I can move them somewhere else.

the growth team, I think that’s where your data team is really responsible for understanding. You should have a suite of metrics that are your growth metrics that can be leveraged by a variety of different teams. And so that to me typically is a data owner or your VP of data, something like that, to be able to provide. Ideally, you want data to not be a reactive part of your team. You want it to be a proactive part of your team. So every week, data should be bringing insights to your product and marketing org site.

There’s a strategy on how we can get more engagement, more value out of each of our customers. Or the opposite being like, hey, I want to work on engagement. Data team, let’s work together to build out a model to help us understand who are the audiences that we should be tackling. Because I, as a marketing team, might want to do that with email right now, or I might want to do that with messaging in the app, or I might wanna do that with an event for our customers.

Whereas product might be, hey, we’re going in and we’re gonna be working on the discovery element of our website. So or our app or ETAs, just think it’ll impact engagement, work into that. So that’s where I feel like data becomes a really great anchor for growth to be infused throughout the organization.

and the way I’ve done it are probably two different things. So the right way is you’re ideally doing them with every sprint, and there’s some type of regularity to them. I’ve worked with a lot of early stage companies. It tends to be that post mortems happen when something breaks. It’s actually like, we should probably figure out what just happened there because something went terribly wrong. But again, good post mortem health would say that you’re actually doing it every time. So you’re also identifying the things that are working and all of those types of things.

A, hey, something went wrong. Let’s do a post mortem. Typically, there’s some really great Miro and Coda. There’s some good just basic templates that you can use. And typically, it’s figuring out who the right stakeholders are. So one example is at Lambda School, we implemented a new chat opportunity for our students and we didn’t loop in customer service. Customer service had an entirely different chat element. It just caused a ton of backlash like across the company because it was like, hey, we got good data out of this, but we totally I like to call it owning your collateral damage.

Everything you do is probably gonna have collateral damage. Go find out what it is. Go embrace those teams. Ideally, you tell them ahead of time. You get their buy in ahead of time and things that. But sometimes I think that’s where postmortems come in handy and say, hey, we misstepped. We’re owning that. We would like you to come in and help us break this down. But doing it with one of those templates and coming in with a, hey, we’re gonna all share in an honest fashion here.

And the goal isn’t to point fingers, but is actually to allow us to do that same thing again, not prevent us from doing that thing again. I think that mindset can really help a post mortem be more effective.

Oh, it ranged. My biggest challenge were the operations teams, because you had every city team would build a small operations team including marketing. At any given time, there were probably 200 personal credit cards buying ads on Facebook, and it used to drive me mad. And initially, I went out with a very firm grip on, you can’t do that. Stop doing that. We have it structured over here. It’s tagged appropriately. You’re bidding against ourselves. A very just authoritarian kind of view on it. And it just never stopped.

Eventually, I shifted that into a, I need to actually build for those teams. Otherwise, they’re gonna keep doing this. And I actually need them. They understand their cities and these dynamics way better than I do because they’re on the ground. And we actually built a function within our team to work directly with cities and that type of stuff. And then I know product and data had those same challenges as well.

Autonomy and go fight to win was very much part of our culture at Uber. If you’re a GM for market and you hire a marketer and you’re like, man, this central team’s doing marketing that’s not quite what we need. A good example are outside lands or Lollapalooza or Coachella, where I as a global marketer, I’m not really building content for Coachella. I don’t really care. That’s not for me. That’s one week. I can buy more efficiently when I think much broader Bay Area. Whereas like for the SF areas, like, our entire market is gonna be screwed for a week.

I wanna go do specific marketing for that. To me as a global buyer, my initial view on that is you’re gonna spend $10,000 of our billion dollar budget and putting three people full time on that, that is a waste of our time. That is not a good use of our time. That was the culture was to go do that because their goal was to get San Francisco to work, to win San Francisco. Eventually, we had to come to terms. It’s just, oh, okay. Maybe I could build things that allow you to do Coachella.

Hey. It looks like every city does a major concert. Maybe we build our own internal templates on how to buy that. So I reduce the noise hitting my team and I improve the effectiveness of what they’re doing at the local level. That’s net good for the business. That’s a hard problem. It’s very hard.

So ultimately, you want to make paid a tool, not a crutch. You want it to be something that you can add fuel when you need that fuel, whether that’s because you’re Zoom and there was a pandemic and you’re like, oh crap. We need to go win the market right now. Move as fast as possible. Just dump all of our money into the next six weeks because it will be more effective now than ever before. In that moment, you should probably be willing to make it 90%.

To have a refined tool, you also don’t want it to be 0% necessarily because you’re gonna miss those moments if you’re not operating it at all. I usually tell people 30% tells me that you’re not dependent on paid, and it tells me that you’re doing it enough to have a need to keep refining and improving it, being relevant in the moment, and those types of things. And by giving that anchor of 30%, it starts to allow people to be uncomfortable when it’s above 30%. It doesn’t mean you shouldn’t go there, but you should start to get uncomfortable.

If it’s like, man, we’re spending way too much money. When you’ve put that anchor at 30, it allows you to at least say how much is too much rather than just saying, oh, this feels off.

So we probably floated around 50% when we were hitting that. So when we were doing a billion dollars, that included China, and that was more than 50%. Paid drove a ton of our growth in China. It was nonsense. It was just wild. Let’s say $6,700,000,000 on, like, the rest of the world to market Uber. That was probably accounting for 40 to 50% of our growth at any given time. Now it’s probably way more efficient than that. They pulled back significantly. We started to their by fourth year at Uber, we started pulling back quite a bit.

Sorry. I I’ve never been to China. What was so wild about it? None of the channels that we’ve used are there. So, oh, we’ve been using Facebook and Google and all these job boards, and none of them exist in China. You literally had to take, like, everything you know about media buying and planning and marketing and throw it out the window, go understand and learn that market, and then figure out how to do all of the different channels there. For instance, the App Store is a much more distributed world there.

It’s not just, hey, drive everything to Apple. It’s, oh, you actually need to go buy relationships with various different App Stores to be able to be there and to be discovered there. When we were able to hire, like, agencies and things like that, it became very clear that one had all our data. So DD knew everything we were doing. Whereas Lyft and Uber always knew what each other were doing because we’re watching each other, we’re friends, things like that. But we never had each other’s data.

China, they definitely had our data. Like, 100%. The Chinese government has it. Beat DD has it. Everybody that wants to compete with us has it. It’s, like, readily in the market for us. Did you have that data? We didn’t. As far as I know, we didn’t. I didn’t have their data. We either have Zoom calls and then just random phone numbers would just show up on our Zoom calls. And we’re like, oh, gotta hang up, get into a new Zoom room, things like that. We’re just, oh, man.

There’s so much it feels like the wild west here at times. But there’s still a corporate structure there that means that you’re not gonna get away with fraud for very long, or that Google and Facebook can’t allow that stuff too much because if they do, it’s gonna hurt their entire business model. And all of that just goes out the window. And then even from the cultural differences with the types of rides, the type we had to do a lot of just different creative, much more loud, crazy characters, that type of stuff worked way better and differently there, a lot more out of home, those types of things.

I agree. I think it’s bullshit. The expectation that it’s gonna be a consistent metric across channels and having a full view of everything that could be included in that is impossible. It’s not unlikely. It’s impossible. Like, it’s impossible to fully attribute all of your channels, let alone, like, one of your channels, especially with the walled gardens of Google and Facebook and Apple and all of this stuff now. I do still think CAC is a very efficient input and tool for a media buyer. But ideally, that CAC is informed by a media mix model that’s constantly you’re basically just trying to optimize for your confidence in your model, not for perfection.

Pulling in more data, pulling in more models, pulling in the information from your partners, all of that stuff, being fed back into the buyers with a we believe that on Facebook, a customer’s value is $20, go get as many as possible, is the best way to let that team operate instead of asking them to decipher all of the magic behind the CAC to LTV.

Yeah. The one thing you just highlighted is that your b to b larger corporate audience, enterprise is a lot easier of a puzzle to solve from the CAC to LTV perspective, whereas a consumer, especially mass consumer marketplace, etcetera, just adds all of this noise and complexity and things like that into it. From your perspective, from a media perspective, makes perfect sense. But from a buying perspective, if I’m a mobile gamer, yeah, you’re right. It’s gonna be hard to attribute it, but I still need those users.

I still need to go buy as many $1 users as humanly possible. Where am I gonna do it? How much so they still have to make that decision. Generally, I think you’re right. Using CAC and just spreading it evenly across the business, CAC, I think is our third trap.

The thing I love about investing is that I because I’m working with the founder with an expectation of how to scale this, a venture scale to this, I get pulled into what I think of as existential problems of the business, not the day to day operational problems. Whereas when I go and I’m consulting or advising, I inevitably get pulled into, hey, what’s the next channel we should be doing and things like that? More often than not, I find that most of the company’s problems aren’t that they haven’t found the channel that’s gonna work for them.

It’s actually that they don’t have product market fit or that they’re pointed the wrong direction or they’re too narrow in how they’ve solved their problem. And so being able to both tie that existential opportunity with the what are the operational asks that they’re giving to their teams right now has just been really rewarding for me, and I think helps me think those big picture thoughts and ground them to like, okay, here’s what to go hand your team to go find what they’re looking for right now.

That’s how it shifted my ability to do both kind of ends of that spectrum.

For me, it’s just become all sports, and I have lots of interests. I like sports, but it just keeps showing me the one thing. I think they’ve over indexed on what they think ads will like versus indexing towards time relevance, which I think is the superpower of Twitter is it’s exactly what’s happening in this moment. And so if I was the growth leader at Twitter, I would be, how do we really capture that magic of this is special right now more than this is special because it’s the generic thing that person likes?

Clubhouse. How do you evaluate what happened? What was the magic and what went wrong? Product market fit on both answers. So I think they had just killer product market fit. The vision initially was how do we create this feeling of a dinner party? Those types of conversations. In that moment, everybody needed that. But very quickly, most people don’t need that every we talked about frequency. Most people want that once a month, not every single day. But they opened it up so quickly to that, and then the product shifted, or the market shifted where that product market fit doesn’t quite fit there.

And this is what a lot of consumer companies are going through right now, where they think that, oh, they gotta keep trying new things, but really, they don’t have product market fit the way they used to in the middle of a bear market

Yeah. I think I think they’ve carved their customers out. I think you’re gonna instill people ebb and flow between them, but just like Snapchat before it. Think But you don’t think they’ll get banned? I don’t think so. I think that there’s too much of a single generation that uses that app that would go crazy if it disappeared.

I’m great at connecting the dots. I’ve had a lot of exposure. We talked about I I studied improv and comedy my whole life, and so I love this idea of yes and connecting pieces. And I find that really effective in the weeds with founders because I’m able to not just think about their singular vision, but really connect it to different consumer experiences, different products. And I think a lot of people get way too pigeonholed into, I have to solve this. It’s a new problem. I can’t learn anything from anyone else.

I gotta go do it myself. Final one. What

I love Epic Games. I think that they’re both building a platform with Unreal Engine and building the best games to be on that platform. Just reminds me a lot of Apple with the App Store early on, and I think we’re starting to see some versions of that with the AI companies hitting the market right now as well. And that to me is the closest to, like, winner take all opportunities, is you building both the platform and some of the best products on it. I’ve actually been, like, really geeking out about some of their strategies within that.