Marketplace Lessons from a Seat on the Uber Rocketship
I joined Uber at the dawn of ridesharing and was part of the rocketship journey of the company to help it become the worldwide presence that it is today. Much of the journey was getting in tussles with local governments, regulators, and politicians as well as launching one of the earliest Uber Eats marketplaces. I learned a lot of lessons through my years at Uber that may be helpful to burgeoning marketplace entrepreneurs.
There has been a lot written about marketplaces. If you are part of a marketplace start-up, or thinking about launching one, there are two ‘stop everything right now and read them resources’ that I consider required reading:
- Bill Gurley’s 10 factors to consider when evaluating Digital Marketplaces
- Version One’s Guide to Marketplaces (edition 2)
While there is a lot of literature out there on marketplaces generally, there really isn’t anything that expresses the learnings I gained over four years at Uber. My Uber experience ingrained a lot of interesting ways of looking at the business, and only after I left and started to look at other marketplaces, was I able to codify some of the tribal knowledge that was ingrained in many of us at Uber.
So I’d like to present to you some additional factors (further to Bill Gurley’s and Version One’s) that I think are important when assessing marketplaces all rooted in the Uber experience:
Marketplace Health
Aside from GMV and take rate, the next most important metric that a marketplace needs to have nailed is related to its overall health. This is usually a good indicator of what the balance of supply to demand is.
As a GM at Uber, one of the most important metrics that I was held to every single week was the Completed / Request (C/R) ratio in my markets. In other words, the number of times a ride was completed divided by the number of times riders pressed ‘Request a ride.’ In the early days of launching our markets, there were several reasons why someone wouldn’t be able to get a ride after pressing the button. This could include i) No cars available, ii) Driver cancels the trip after not being able to find the rider (or other reason) iii) Rider cancels the ride because surge was too high. This ratio was a fast metric that told us how healthy a market was, if C/R was too low, we knew that supply was the constraint, if it was too high, demand was. We set a goal at a specific C/R%, which was a tipping point where riders saw increasing value, and therefore we saw compounded interest in demand — meaning, if riders saw the reliability in the system was there (if they requested a ride, they were able to get one) they actually started to use it more and more.
C/R will surely not be the health metric that every marketplace should be using, but there should be one critical metric that really speaks to the health of that marketplace. I start to get excited in a marketplace opportunity when the founders show me they know what this is for them and what the levers are to pull to improve this core metric.
Further, having this marketplace health metric at the most granular level possible allows your team to easily discern where priorities should lie. At Uber, we could see this metric as granular as the neighborhood and by time. Granularity helps translate metrics into action e.g., “I need to focus on building driver supply in downtown, at 6PM on weekends.”
Supply Friction Mitigation
Bill Gurley lists ‘Friction of Supplier Sign-up’ as one of the 10 factors to consider when evaluating marketplaces. I agree with him, all else being equal supplier friction is bad and if possible should be avoided at all costs, but I would add a nuance to his premise.
Structural friction is OK if the company has figured out a way to operationalize out as much of that friction as possible because it can create a moat against competition.
Good Uber examples would be ridesharing in New York City or Houston. Historically, the regulations in both cities were a lot more onerous; to become a taxi driver or a ridesharing driver it would be a multi-week or multi-month process. In both situations, Uber took the step of operationalizing out as much of that friction as possible, (e.g., in the case of Houston, ‘Become an Uber driver in a day’). Where their hubs became all-in one-shots, where a driver could get a medical, take applicable training classes, and undergo vehicle testing all at once. Having these high barriers for entry (but operationalized simplicity) kept more of the competition at bay and allowed Uber to take more of the market in a competition constrained environment.
First to Scale From the Competition?
First to market in your segment doesn’t matter, however first to scale does. I get extremely excited about an investment opportunity when they show me that Product-Market Fit in that one vertical / segment or geography, and it gets even more exciting if they can show me they’ve figured out the playbook to expand it to one more vertical or geography. The ability to playbook and replicate is not an easy task, but there’s a step change in my interest if you can prove it to me just once. This also goes to say, start small (whether that’s one sector or vertical or neighborhood) and go deep rather than a mile wide and inch deep.
A lot of people forget that Uber wasn’t the first to market for ridesharing. A fast history lesson: UberBlack was Uber’s original product, which was all limousines and towncars, and was definitely an upscale experience and price-point. Lyft and Sidecar actually came out with “everyday” ridesharing first. Uber waited for 9 months on the sidelines until Travis Kalanick (then CEO) put out a whitepaper saying Uber was entering the ridesharing market. There were a lot of factors that contributed to the velocity to scale that Uber had relative to competitors, including an international decentralized structure that allowed for hundreds of cities testing new local strategies in parallel, and adopting best practices rapidly. These learnings along with learnings from scaling UberBlack were an insurmountable force compared to the competitors at the time, this is in large part why Uber is the Kleenex of ridesharing today. They weren’t first, but they sure scaled first.
Latent or Underutilized Supply
There’s countless articles written on the web in regards to what a marketplace company should focus on first when starting out — supply or demand? The overwhelming advice is that supply should be your first focus. The same was true for Uber. When we would first launch a city, we would pay drivers a strong per hour rate just to be out there driving, irrespective if they got trips or not. “If you build it, they will come,” was a common phrase thrown around. In the early days you want to avoid marketplace failures (a potential buyer comes and there’s nothing to buy) as much as possible, this ensured the flywheel of demand virality was able to take off quickly.
More importantly, I ascribe a lot of Uber’s success to the ability to ‘hack’ the early marketplace dynamics by leveraging latent or underutilized supply. In the first incarnation of Uber — UberBlack, the team realized that limousine / town car drivers were busy during the mornings, taking trips to the airport, and busy again in the late afternoons with trips back from the airport, but generally unutilized during the day. Supply was readily available driver willingness was high since they could earn extra income while they were free anyway.
When you think about uberX and ridesharing, the same theory applies. Cars sit parked in parking lots 95% of the time. If people are looking for work and they own a car, this is a latent supply source that can quickly ramp up.
Lastly, take the success of UberEats, this is Uber using their own latent or underutilized supply. In this case, uberX drivers may not be busy all the time, why not have them make more money by delivering food in between delivering people.
This latent supply goes a long way to helping create your first power-users e.g., in Uber’s case, the drivers that are willing to drive fulltime and resulting in early marketplace liquidity. Companies that find ways to leverage underutilized supply (supply that does not require a lot of friction / effort to become qualified supply) build liquid marketplaces faster.
Ability to Become Full-Stack / Incorporate Adjacencies
This last point is one I believe applies only to later stage marketplaces, and it is one that I caution early-stage founders to spend too much bandwidth against. Referencing back to First to Scale from Competiton, it’s crucial that marketplaces focus their efforts into creating the magic in that one very specific vertical or sector or geography first.
For those later stage marketplaces, can they plausibly take on category adjacencies and own those marketplaces as well?
Uber’s example here is quite direct. When Uber first started it was almost always the case that only ridesharing or the original UberBlack were available on its platform. This allowed for the Uber experience to be uniform and reliability to be high. With few exceptions, Uber steered away from having other modalities like taxi on the platform.
More recently what you see is Uber is looking to become the ‘Amazon of Transportation,’ where they are looking to handle all forms of transportation irrespective of modality. This is of course only possible after a certain level of scale and recognition. This is what has fueled the acquisition of Jump, and partnerships with Lime, and Taxi, and increasingly with public transit offerings. Since Uber was able to own a significant percentage of how people got around, it’s entirely plausible that all forms of getting around could go into the Uber marketplace.
So, putting these above factors together with Version One’s marketplace scoring framework (which I find is a little more current for venture scale marketplaces than Bill Gurley’s from 2012), here’s our ‘work in progress’ Inovia Capital marketplace scoring framework:
This is the first framework we use to determine how much risk is inherent in a marketplace, a higher score means less risk. It’s not necessary that a low score is necessarily bad, but may require more digging to ultimately justify an investment.
As you are thinking through your marketplace, this may be a helpful tool to evaluate approaches to be taken.
I fully expect this will continue to be a living framework and will be updated periodically, but in the spirit of getting minimum viable products out to market as soon as possible, here you go, world.
— Ramit Kar
Ramit was a General Manager at Uber during the launch of ridesharing, after spending over 3 years launching and scaling ridesharing and Eats businesses, he is now an Entrepreneur in Residence at Inovia Capital
