Uber: la carrera por la liquidez en marketplaces de dos lados | Blog SMX

Quick summary

Uber did not win because it had a better app. It won because Travis Kalanick understood something most founders never internalize: in a two-sided marketplace, the scarce resource is liquidity — not capital, not marketing, not technology. Every one of the four decisions that defined Uber between 2010 and 2014 — launching at SXSW instead of NYC, subsidizing drivers not riders, saturating one neighborhood before covering a city, and hiring street teams instead of buying Facebook ads — was a bet on the same thesis: seed the constrained side first, compound density faster than your competitor, and let the network effect do the rest.

What is Uber, exactly?

Uber — originally "UberCab" — is a ride-hailing platform that connects passengers requesting a trip with drivers willing to provide one, in real time, through a mobile app that handles routing, pricing and payment. It was founded in March 2009 by Garrett Camp and Travis Kalanick in San Francisco, and launched its first commercial service in mid-2010.

Mechanically the product is simple: a rider taps a button, the app shows nearby cars on a map, matches the rider with the closest driver, routes the vehicle via GPS, and charges the rider's saved payment method at the end of the trip. No cash changes hands. No phone call to a dispatcher. No uncertainty about whether a car is coming.

Seen from the outside, Uber looks like a software company. It is not. Uber is a two-sided marketplace — a platform where two distinct user groups (riders and drivers) derive value from each other, and where that value grows as the number of users on each side grows. The app is the interface. The real product is the match. And the real strategic challenge is making sure that match happens fast enough, often enough, that both sides stay on the platform.

Economists Jean-Charles Rochet and Jean Tirole published the foundational paper on this class of business in 2003 ("Platform Competition in Two-Sided Markets"), and Thomas Eisenmann, Geoffrey Parker and Marshall Van Alstyne translated it for managers in their 2006 Harvard Business Review essay "Strategies for Two-Sided Markets". If you read those two pieces you can derive every single thing Uber did in its first five years from first principles. Kalanick did not read them before launching Uber — but he was acting on the same logic.

The context: why 2010 San Francisco?

To understand why the Uber story starts where it does, you have to understand the San Francisco taxi market of 2010. It was, by common agreement, the worst major taxi market in the United States.

San Francisco had roughly one-tenth the taxi density per capita of New York City. The city had capped medallions since the 1970s and rarely issued new ones. Dispatch systems were radio-based, run by half a dozen fragmented companies, and notoriously unreliable — riders routinely waited 20 to 45 minutes for a cab that might never arrive. Credit card readers, when they existed at all, were usually "broken" when the meter hit double digits. The whole experience oozed contempt for the customer.

Inside this broken market, Garrett Camp had a prototype idea: what if, instead of calling a yellow cab, you pressed a button on your phone and a black town car showed up in five minutes? Town cars — the ones that wealthier people already used for airport runs — had plenty of unused capacity between scheduled jobs. The hardware existed. The drivers existed. The only thing missing was coordination.

Kalanick, the son of a Los Angeles civil engineer, was 33 and carrying two startup scars. His first company, Scour, had been sued into oblivion by 29 Hollywood studios. His second, Red Swoosh, had been a slow-motion grind that ended in a US$ 19M sale to Akamai. He had burned out, traveled, and come home broke and hungry. Camp offered him the CEO role at UberCab. He took it.

The San Francisco launch worked almost immediately. Early users — mostly tech workers who already had iPhones, credit cards and disposable income — became evangelists. The seed round of US$ 1.3M led by First Round Capital, followed by another round bringing working capital to about US$ 2M, gave Uber the balance sheet of a company with maybe 14 months of runway. Not much, but enough to try something.

And something had to happen quickly. A Stanford-educated Santa Barbara founder named Logan Green was circling the same space with a carpooling app called Zimride that would soon pivot into Lyft. Other ride apps were launching in parallel. Every week Uber spent perfecting San Francisco was a week a copycat could use to plant a flag in another city and make themselves the category-defining brand there.

~1/10 — SF taxi density vs. NYC per capita

US$ 2M — Uber working capital in late 2010

1 — Cities live at start of 2011

<5 min — SF wait time for an Uber (vs. 20+ for a cab)

The 4 decisions that defined Uber

There are several ways to tell the Uber story. The way I find most useful — and it is how we structure it in the SMX simulator — is as a sequence of four strategic decisions, each with plausible alternatives Kalanick could have taken. In each one, the "sensible" answer differed from what Uber actually did. And in each one, the difference between Uber and its graveyard of competitors traces back to the decision Kalanick made.

Decision 1: Which city goes second — Austin, New York or Boston?

By late 2011, Uber had to pick its second city. The team had three memos on the table. Austin during SXSW would spend roughly US$ 250K to flood the festival with free black-car rides — a PR detonation aimed at the journalists, venture capitalists and early-adopter founders who all converged on the same square mile for one week in March. New York was the biggest taxi market on Earth but also the hardest regulator (the Taxi & Limousine Commission) and the place where unit economics would be tested under maximum pressure. Boston was a controlled, mid-sized market — cheap to run, manageable regulator, perfect for polishing a replicable playbook before touching a giant.

The "sensible" MBA answer is Boston. You learn cheaply, you de-risk, you build an operations manual. The slightly braver answer is New York — prove you can survive the hardest market and everything else becomes easier. Both answers miss the point.

Kalanick chose Austin at SXSW 2012 — and it was the right answer, not because SXSW was cheap (it wasn't) or because Austin was a big market (it wasn't), but because narrative control is a weapon in a land-grab market. The SXSW stunt generated over 200 press mentions. Every tech founder, journalist and VC in America knew the word "Uber" by the end of the week. The Series A that followed was the easiest fundraise in recent memory. More importantly, the stunt built a myth that made every subsequent city easier: when the Uber launch team showed up in Chicago, Boston or Philadelphia, drivers already knew the brand and riders had already read about it.

The counterfactual is instructive. If Uber had picked Boston, it would have been competent and invisible. If it had picked NYC, it would have produced clean unit economics but ceded the narrative to Lyft — and "the Boston company that also works in San Francisco" is a ceiling you cannot escape in a two-sided marketplace race. The surgical answer was wrong because the market was not surgical; it was a race, and the prize went to whoever owned the story first.

Decision 2: Who do you seed first — drivers or riders?

In every new city, Uber faced the same question, and it is the question every two-sided marketplace founder faces in every category. You arrive with zero drivers and zero riders. You have a budget. You can subsidize one side heavily, both sides moderately, or split the difference. Which do you choose?

The spreadsheet-friendly answer is "a little of both". The growth-hacker answer is "riders, because referrals are viral and cheap". The operational answer is "drivers, because supply is the constrained side". Uber picked the third one — and picked it violently. Early drivers in new cities received hourly guarantees of US$ 25–30 whether or not they got rides. The subsidies were frequently the single largest line item in a new city's operating budget for the first four months, and they tapered only as organic demand caught up.

Why? Because supply is patient and demand is not. A driver who sits empty for an hour logs off and tries again tomorrow. A rider who opens the app, sees no cars and waits 20 minutes deletes the app and tells three friends not to bother. The first rider experience is the single most valuable asset a marketplace builds, and you only get one shot at it per user. Subsidizing drivers means that first experience is always magical: press button, car arrives in four minutes, no friction. That magic converts to a 70%+ repeat rate. Referrals without liquidity convert to a leaky bucket.

The Rochet-Tirole formalism behind this is worth stating plainly: in a two-sided network, one side is always the "money-losing side" in the cold start and the other is the "money side". Your job as an operator is to figure out which is which — and for physical marketplaces with human supply, the supply side is almost always the one you subsidize. Uber understood this in 2010. Most of its dead competitors did not.

Decision 3: How do you cover a city — hyperlocal density or citywide breadth?

Once Uber launched a city, the next question was how to cover it. One internal faction wanted hyperlocal dominance — pick one dense neighborhood, saturate it with drivers until pickups were consistently under five minutes, and only then expand outward. Another wanted a citywide land grab — launch all five boroughs (in NYC) or all central zones (everywhere else) simultaneously to deny the competitor any foothold. A third argued for premium-only — stay black-car, high margin, never dilute the brand.

Uber's playbook — eventually codified and industrialized across more than 700 cities — was hyperlocal dominance. A "launch team" of three operations people would parachute into a city, identify the densest possible zone (usually the financial district combined with the nightlife corridor), and concentrate every driver guarantee, every rider promotion and every event activation inside that zone for roughly six to eight weeks. Only after wait times inside the zone were averaging under four minutes would they ripple outward.

The reasoning is that density compounds; breadth churns. Five drivers spread across the five boroughs of New York deliver a 45-minute wait to everyone; five drivers concentrated in lower Manhattan deliver a two-minute wait to a subset — and that subset tells three friends, each of whom tells three friends. Dense liquidity produces a viral surface. Thin liquidity produces cancelled rides, one-star ratings and a churned user base that is permanently skeptical.

The premium-only path deserves one sentence of respect: it was genuinely tempting in 2012, and some investors pushed for it. The reason Uber rejected it is that in 2012 they launched UberX — a mass-market tier that cut ride cost by roughly 60% and multiplied the addressable market tenfold. Every subsequent competitor lesson confirmed it: premium-only is a survivable niche, but it is not a platform.

Decision 4: How do you buy the city — digital ads, street teams or brand?

CAC rises in every city over time. The cheap early adopters saturate; competitors start bidding on the same keywords; cost per rider doubles within four months. The question is where the next marketing dollar goes.

The technocratic answer is digital ads at scale — Facebook, Google and Instagram give you precise targeting and measurable CAC. The executive answer is brand advertising — billboards and TV build long-term awareness. Uber chose neither. Uber bet on street teams and local partnerships: launch crews stationed at airports during Friday-night storms, partnerships with law firms and consulting firms for corporate accounts, sponsorships at music festivals and tech conferences, and free-ride codes handed out physically at nightclubs and bars.

The numbers were striking. A rider acquired at JFK during a rainy Friday rush cost roughly US$ 20 to acquire and converted at over 60% on the first ride. A rider acquired via Facebook cost US$ 12 initially but US$ 45 by month four, and converted at 20%. More importantly, event-acquired riders referred an average of three friends. Facebook-acquired riders referred almost none. The reason is simple: getting into an Uber for the first time at 11pm, in the rain, after watching a launch-team employee high-five the driver, produces an emotional bookmark that no display ad can match.

This does not mean Uber never used digital ads or brand campaigns. It means Uber used them at the right stage. Digital ads came in force once liquidity was dense enough in a city that the ads amplified a working experience rather than funneled users to an empty app. Brand advertising came after 2017, when the company needed to repair trust after a series of governance crises. Both tools work — but only as follow-on leverage on an already-liquid marketplace, never as the primary engine of cold-start growth.

What Uber has become

The scale of what Uber became between 2010 and 2024 is hard to process without losing proportion.

700+ — Cities with Uber operations

150M+ — Monthly active platform consumers (2024)

US$ 82B — IPO valuation (2019)

US$ 3.6B — Cumulative rider/driver subsidies 2010–2019

Read in isolation, that US$ 3.6 billion in cumulative subsidies looks like spectacular waste. Read through the lens of two-sided marketplace theory, it is one of the best-targeted capital deployments in startup history. Every subsidy dollar bought a week of compounding liquidity in a specific city — a week that Lyft, Hailo, Sidecar, Gett or any of a hundred dead competitors could not recover. The subsidies were not an expense. They were the entry fee for a permanent monopoly-or-duopoly position in the single largest category created by the smartphone.

The Uber launch playbook became an industrial process. Three operations people, six weeks, a predictable budget, a standard set of partnerships to close, a standard set of neighborhoods to saturate. The playbook was replicated in every major city in North America, then Europe, then Latin America, then Asia, with only cosmetic adaptations. When you see a team from DoorDash, Rappi, iFood or Gojek launching a new city today, you are watching the Uber playbook executed in a different category.

The economic impact is enormous but also contested. Uber generated millions of income-earning opportunities for drivers — a complement to traditional employment that many riders and drivers value. It also absorbed a meaningful share of what would otherwise have been taxi, car-ownership or even public-transit spending, and it reshaped urban labor markets in ways that regulators, academics and the drivers themselves are still arguing about. The story of Uber is not morally simple. But as a case study in how to scale a two-sided marketplace, it is unsurpassed.

Want to live the decision, not just read about it? In the SMX simulator you sit in Travis Kalanick's chair in late 2010, with US$ 2M in the bank and one live city. You make the four decisions in this article in 30 minutes, see the financial and strategic consequences of each choice, and get scored against what Uber actually did. Strategic feedback, comparison with the real decision and certificate included. Try the Uber case in the simulator →

3 lessons that travel beyond Uber

It is tempting to read Uber as a story about transportation, or about a single combative CEO. It is neither. The principles behind the four decisions show up, with different costumes, in every major two-sided marketplace that has scaled since — from DoorDash to Airbnb to Substack to OnlyFans to Upwork. Three are worth stating plainly.

1. Liquidity is the only KPI that matters in the first 24 months

Signups are vanity. GMV is vanity. Website visits are vanity. Liquidity — the probability that a user's request is fulfilled in a short, predictable window — is the only metric correlated with long-term survival in a marketplace business. If your liquidity metric is going up, your business is alive. If it is flat or declining, no amount of paid marketing will save you; you are acquiring users into a product that does not yet work, and each of them is a future one-star review. Every dashboard, every board slide, every OKR for the first two years of a marketplace should be built around a single question: did liquidity on the constrained side improve this week?

2. The constrained side must be subsidized violently, not politely

Balanced subsidies are the single most common killer of two-sided marketplaces. They feel prudent on a spreadsheet — you hedge, you don't over-commit, you're fair to both sides. In practice, balanced subsidies mean neither side reaches critical mass, you burn cash in two directions, and six months later you have no liquidity and half the runway. The right answer is brutally asymmetric: identify which side is structurally scarce in your category (drivers for ride-hailing, couriers for food delivery, hosts for short-term rentals, creators for content platforms), and over-invest in that side until the other side starts self-balancing. Politeness is not a strategy.

3. Density beats breadth in every cold-start market

Every marketplace founder is tempted, at some point, to "go nationwide" or "launch ten cities at once". The temptation usually peaks right after a fundraise. The counsel from the Uber playbook is blunt: do not do this. Ten cities at 10% liquidity each produce ten one-star first experiences and a decade of reputational damage. One city at 100% liquidity produces a playbook you can replicate. The correct sequence is always: saturate, replicate, scale — in that order, never reversed. This is equally true for food delivery within a metro, for Airbnb within a neighborhood, for a creator platform within a content niche. Dense liquidity compounds; broad mediocrity churns.

Conclusion: what Uber teaches about strategy

If you read the classical competitive strategy books — Porter, the Five Forces, SWOT analysis — none of them cleanly predict a case like Uber. That is not a failure of theory; it is a reminder that strategy, in practice, is a sequence of decisions made under uncertainty, where the "right" answer only becomes obvious in retrospect. Kalanick did not know in 2011 that SXSW was the right second city. He made a bet, under time pressure, with incomplete information. The same is true for every decision that followed.

The real value of studying cases like Uber is not memorizing the SXSW stunt or the driver guarantee playbook. It is training the muscle of making real strategic decisions — with incomplete information, reluctant counterparts, brutal trade-offs, and windows of opportunity that close. You do not become a good CEO by reading about CEOs. You become a good CEO by making hard decisions and receiving honest feedback on them.

That is exactly what we have built at SMX.