why Uber launched with black cars — and how the pattern is running inside Microsoft right now

Filed under: Two-sided marketplaces · Cold-start problems · Growth strategy classics · Enterprise growth

There's a story in Uber's early history. I think about it nearly every day. Because it maps onto a growth motion that runs at enterprise scale, with a very different-looking supply side.

When Uber launched in San Francisco in 2010, it didn't launch as the app we know today. It didn't launch with regular drivers using their own cars. It didn't undercut taxis on price. It launched as UberCab, a black-car service, priced at roughly 1.5x the cost of a taxi, targeting exactly the customer already willing to pay a premium for a private car.

The consumer-facing story of Uber — on-demand rideshare with a phone, an army of everyday drivers, pricing pressure on incumbents — didn't exist yet. That version only became possible in 2013, after Uber launched UberX with regular drivers and personal vehicles, and only after Lyft, Sidecar, and Wingz had already fought the regulatory battles that made peer-to-peer rideshare legal.

For roughly two and a half years, Uber was a nicer black-car dispatch service. That was the entire product.

Why? Because the hard part of any two-sided marketplace isn't the demand side. It's the supply side. And if you're going to solve supply, you solve it in the corner of the market where supply is already licensed, insured, professional, and — critically — sitting idle between pre-booked rides.

This is the growth strategy lesson every PM building a two-sided motion should study. The framing that gets Uber wrong is "start luxury, go downmarket." That's a marketing story about positioning. The real story: cold-start where the supply is already latent, price high enough to compensate that supply for their time, and don't try to change the shape of the market until the loop works on easy mode.

The demand-side myth

The obvious way to launch a rideshare product is to think about riders. Riders are the fun side. Riders are the ones with the visible pain (waiting for a cab in the rain). Riders are the ones with the credit card. Every consumer PM starts on the demand side because demand is where the visible market failure lives.

But demand-side launches fail. Not always, but reliably, in two-sided markets. The reason is simple: a good experience can't get delivered if there's no supply. A rider opens the app, doesn't see a car within five minutes, closes it, doesn't open it again for six months. That's a lifetime value burned in the first thirty seconds. Multiply it across a launch city and the market's burned for anyone, including the company that burned it, later.

The only way to launch a two-sided motion is to solve supply first. And the only way to solve supply first is to find supply that doesn't require changing the world to activate it.

Where the black cars came from

Black-car drivers in every major U.S. city were, in 2010, a professionalized supply base with a specific structural problem. Licensed, insured, commercially permitted. Capital tied up in vehicles. Scheduled rides — a hotel pickup at 6 a.m., a corporate account at 4 p.m. — but between those rides, sitting in an airport parking lot, meter off, waiting.

That idle capacity was the entire opportunity. Uber didn't have to convince anyone to become a driver. Didn't have to solve licensing, insurance, or vehicle procurement. Didn't have to fight the regulatory battle. Had to solve exactly one problem: give black-car drivers a way to fill their empty hours at a price that made it worth their time.

The mechanics are worth sitting with, because the specificity is the whole lesson, and it's a more extreme lesson than "incentivize the supply side" makes it sound. Uber paid licensed limousine and town-car drivers a guaranteed hourly rate — often $30 to $40 or more — just to stay logged into the app and circle the geographic core of a city. Guaranteed. Whether they booked a single passenger that hour or none at all. That's not a discount, a bonus, or a referral credit. That's real wages, driver by driver, hour by hour, for supply that might not convert to a single dollar of revenue — on purpose, before a single rider had proven the product would even work.

Call it what it is: an almost reckless level of commitment to manufacturing supply out of nothing. Uber decided the map could never show an empty city to a new rider, and was willing to burn cash on driver wages with zero revenue attached to make that true. Every other cold-start tactic in this post is a variation on "find latent supply and give it a tool." This one is a level past that — buying the appearance of liquidity before the liquidity is real, at a price calibrated to a driver's full opportunity cost, not a rider's willingness to pay.

The recruiting matched the mechanics. No broad advertising campaign — Uber's early operations people walked into airport parking lots and limousine holding areas and talked to drivers directly, pitching the app as a way to monetize the dead time between scheduled corporate pickups. One driver at a time, in the exact physical location where the idle supply was already sitting.

The premium rider price wasn't a positioning move. It was a supply activation mechanism. The 1.5x-taxi price point was calibrated to the black-car driver's opportunity cost, not the rider's willingness to pay. Riders were, in effect, subsidizing the supply side into existence. Uber then built a demand base habituated to on-demand black cars, and that habit — "a car with my phone in five minutes" — became the wedge that let the company later disrupt the mass-market taxi business with UberX.

Cold start solved. Habit built. Then, and only then, the price came down.

A different story, worth separating out. The early supply subsidy sometimes gets conflated with two much later, unrelated controversies. In 2015, researchers noted that the moving car icons on the passenger map weren't strictly real-time driver positions — closer to a signal of availability than a live feed, which Uber disputed but never fully explained. And during Uber's more aggressive expansion years, an internal tool called Greyball served spoofed, ghost-car versions of the app to code-enforcement officers trying to sting the company in cities where it was operating without a permit. Neither has anything to do with the 2010 cold start — one's a mapping dispute, the other's a much later regulatory-evasion tool — but they're worth naming plainly so the actual playbook (real drivers, a real subsidy, real one-at-a-time recruiting) doesn't get mistaken for something synthetic.

What Kalanick actually believed

The most interesting wrinkle in the Uber story: Travis Kalanick himself, at the time, didn't believe a rideshare model with regular drivers and personal vehicles would work. He said publicly that the first peer-to-peer rideshare company was illegal and would fail. He believed the black-car model was the product.

He was wrong about that. But he was right about the sequencing. UberX only became viable after competitors absorbed the regulatory blows and proved the mass-market model would fly. Uber then moved into that market with a habituated demand base, a scaled supply-activation playbook, and enough operational muscle to run over the incumbents.

The lesson isn't that Kalanick had the future mapped out. It's that the constraint-respecting version of his product got him to a position from which the constraint-breaking version became possible. He didn't solve everything at once. He solved supply on easy mode, built the loop, and rode a broader market shift he didn't personally cause.

The pattern shows up everywhere

The Uber cold start is the archetype, but the pattern repeats:

• Airbnb solved supply first — not by convincing homeowners to rent to strangers, but by finding a supply base (Craigslist listers) already trying to rent to strangers and giving them a better tool.
• YouTube solved supply first — not by convincing amateurs to shoot video, but by tapping the enormous pool of people who'd already shot video and had nowhere good to host it.
• OpenTable solved supply first — not by convincing restaurants to accept reservations, but by giving restaurants that already took reservations a better back-end and eventually surfacing them to diners.
• Stripe solved supply first — not by teaching developers about payments, but by giving developers already trying to build payment flows a dramatically easier API.

In every case, the winning move was the same: find the supply that's already there but latent, activate it with a tool that solves their existing problem, let the demand flywheel form on top. Call the pattern latent supply activation. Not "build both sides at once." Not "start with demand and hope supply follows." Find the corner of the market where supply is already sitting idle, waiting for a better tool — and build that tool.

The enterprise version

growthstack isn't a consumer-marketplace blog. So the fair question: what does the Uber cold-start pattern have to do with running a growth motion inside a large enterprise software company?

Everything, actually. Enterprise growth is also a two-sided problem — the sides are just less obvious.

Inside Microsoft, on the connectivity surface this whole series has been about, the "supply side" of adoption is the field account team — the customer success architects, technical account managers, and sellers who own the relationship with the largest enterprise customers on the planet. They're the black cars of enterprise growth. They already have the relationships. They already know their customers' architectures. They already have the mandate to recommend platform moves. What they don't have is the fifteen minutes a week it takes to sift joined telemetry, verify the pattern, and craft a personalized message that actually lands.

The mirroring engine from the first growthstack post is, structurally, a latent-supply activation tool. It gives the field team a joined view they couldn't produce on their own, a personalized draft they didn't have time to write, a specific customer signal they could act on in an afternoon. The engine doesn't replace them. It gives them a better tool for the job they were already trying to do. Supply activates. Demand — enterprise customers who suddenly hear a coherent, well-informed pitch from their trusted account owner — flywheels on top.

This is the enterprise version of Kalanick's black cars. Find the supply that's already partway there. Give it a better tool. Don't try to change the shape of the market until the loop works on easy mode.

And then — the constraint-breaking version

The real reason the Uber story keeps coming back up is what happens after the black-car phase. Kalanick didn't stop at UberCab. He used the loop he'd built to activate a much wider supply base — regular drivers with personal vehicles — the moment the market shifted enough to make that viable.

The equivalent move inside enterprise is where the growth stack actually gets interesting. Once the field-team loop is running — and it is running — the next question is: how does a much wider supply base of advocates get activated? Not just the paid account team, but a broader community of technical champions across customers, partners, and internal cross-functional roles who have the pattern-match instinct and the audience but no formal mandate. The equivalent of UberX.

That's the design work behind two paired growth-program proposals in flight right now. One activates a broad community of technical champions — the ambassadors — with better tools and formal recognition. The other institutionalizes the underlying competence with a credential. Together they move supply activation from "the field team already in place" to "everyone in the ecosystem who could plausibly become supply." Same cold-start pattern. Wider supply base. New loop.

Black car first. UberX later. But UberX only happens because UberCab happened first.

What every PM building a two-sided motion should take from this

Three things.

One: solve supply first, but be honest about which side is supply. In consumer marketplaces, supply is drivers, hosts, sellers. In enterprise, supply might be account teams, champions, or developers. Whichever side of the motion is scarce and hard to activate — that's supply. Solve it first.

Two: activate supply where the constraint is already partially broken. Black cars were licensed, insured, and idle. Craigslist listers were already renting to strangers. Enterprise account teams already have the customer relationships. The pattern is finding the corner of the market where activation is a tool problem, not a persuasion problem. Persuasion-first launches burn the runway. Tool-first launches compound.

Three: the constraint-respecting version earns the right to build the constraint-breaking version. Uber didn't launch with regular drivers, but launching with black cars is what made regular drivers eventually possible. The first product doesn't have to be the final product. It has to be the product that earns the position to build the final one. Sequence is the strategy.

The lesson isn't about black cars. It's the discipline of not trying to solve every side of the market on day one, the humility to launch the version that can actually work, and the strategic patience to keep a clear map of the version that eventually gets built.

That's the whole game.

Next growthstack post: back to internal signal — turning the customer-journey view into a churn-risk score, and why the customer-issue layer is the single best retention asset most PMs are ignoring.

— growthstack