The Ghost Fleet
In the spring of 2019, roughly 14,000 white-and-blue Smart ForTwos and Mercedes-Benz CLA sedans sat scattered across the streets of eight North American cities, each one a €15,000 sensor-laden bet that urbanites would abandon car ownership for the privilege of paying 41 cents per minute to drive someone else's property. By the end of that year, every single vehicle would be gone — pulled from curbs, loaded onto flatbeds, and liquidated — as car2go, the world's largest free-floating carsharing service, collapsed its North American operations in a controlled demolition that its parent company, Daimler AG, framed not as failure but as "strategic refocusing." The phrase did a lot of work. What it concealed was a decade-long experiment in which one of the world's oldest automakers had attempted to reinvent its relationship with the automobile itself, spending an estimated €500 million to prove that the future of urban mobility was access, not ownership — only to discover that the unit economics of floating a fleet of depreciating assets across sprawling cities, without dedicated parking infrastructure, without density sufficient to generate utilization rates above 15%, and without a pricing model that could simultaneously attract riders and cover the staggering costs of insurance, maintenance, relocation, and municipal permitting, were not merely difficult but perhaps structurally impossible at the scale Daimler envisioned.
Car2go is not, in the conventional sense, a success story. It is something more instructive: a precisely documented case study in what happens when a legacy industrial corporation attempts to execute a platform-era business model using industrial-era assumptions about capital deployment, customer acquisition, and competitive moats. The company's trajectory — from a single pilot of 50 Smart cars in Ulm, Germany, in 2008, to a peak fleet of more than 14,000 vehicles across 26 cities on three continents, to its absorption into the SHARE NOW joint venture with BMW and subsequent retreat from most markets by 2020 — traces the full arc of an idea that was strategically correct and operationally ruinous.
By the Numbers
Car2go at Its Zenith
~14,000Vehicles in fleet at peak (2018)
26Cities across North America, Europe, and Asia
3M+Registered members worldwide
$0.41/minPeak per-minute rate in U.S. markets
~$500MEstimated cumulative investment by Daimler
2008–2019Operational lifespan (pilot to North American exit)
~15%Estimated average fleet utilization rate
A Carmaker's Existential Wager
The logic was seductive, and in 2007, when Daimler's research division first began modeling the concept, it was also genuinely novel. The automobile industry had operated for over a century on a single transactional premise: manufacture a vehicle, sell it to an individual, recognize the revenue, move on. The customer bore all subsequent costs — depreciation, insurance, fuel, maintenance, parking — and the manufacturer's relationship with the product effectively ended at the dealership door. Car2go proposed to invert this entirely. Daimler would retain ownership of a fleet of its smallest, cheapest vehicles, distribute them across a city's public streets, and allow registered users to locate, unlock, and drive any available car using a smartphone app, paying only for the minutes they used. No reservations required. No fixed pickup or return locations. No membership fees in most markets. Just open the app, find the nearest blue dot, and go.
The concept was called "free-floating carsharing," and it was meaningfully different from the station-based models that Zipcar had been operating in North America since 2000. Zipcar required users to reserve a specific vehicle at a specific location and return it to that same spot. Car2go eliminated the round-trip constraint — you could pick up a car near your apartment, drive it to your office, park it on any legal street space within the service area, and walk away. The next user might take it somewhere entirely different. In theory, this created a self-organizing transportation network, a kind of distributed fleet intelligence where vehicles migrated toward demand through the collective behavior of users.
In practice, it created a logistics nightmare of staggering complexity.
We are evolving from an automaker to a mobility company. We will not only build the best cars in the world — we will also provide the best mobility services.
Dieter Zetsche, Daimler's charismatic, mustachioed CEO from 2006 to 2019, became the most prominent evangelist of this transformation. A mechanical engineer by training who had made his name restructuring Chrysler during the catastrophic DaimlerChrysler merger, Zetsche understood better than most that the traditional auto business was a margin trap: enormous capital expenditure, cyclical demand, regulatory pressure on emissions, and a creeping awareness that younger urban consumers — the growth demographic every premium brand needed — were delaying or declining car purchases entirely. McKinsey, BCG, and every consultancy with a transportation practice had published the same chart: vehicle miles traveled per capita in the U.S. had plateaued, urban populations were growing, and the smartphone had created a coordination layer that made shared assets suddenly viable. The question was whether an automaker could capture this shift or would be captured by it.
Fifty Smart Cars in Ulm
The origin was modest enough to be charming. In October 2008, car2go launched a pilot program in Ulm, a mid-sized city of 120,000 in the German state of Baden-Württemberg — conveniently, the backyard of Daimler's Stuttgart headquarters. Fifty white Smart ForTwo microcars, each barely eight feet long, were parked on city streets and made available to 200 pre-registered test users. The pilot was managed by Daimler's Business Innovation unit, a small team operating with startup pretensions inside a corporate structure that employed 272,000 people and generated €95.8 billion in revenue that year.
The Smart ForTwo was the obvious fleet vehicle. Daimler had been manufacturing the diminutive two-seater since 1998, originally as a joint venture with the Swiss watchmaker Swatch — an unlikely partnership whose origin story belongs to the annals of corporate ambition outrunning consumer readiness. The car was cheap to produce, nimble in urban environments, and could fit into parking spaces that would defeat a standard sedan. Its limitations — minimal cargo space, no backseat, a driving experience that ranged from "adequate" to "genuinely frightening on highways" — were irrelevant in a carsharing context where the average trip was 20 minutes and covered fewer than six miles.
The Ulm pilot confirmed the basic demand thesis. Users adopted the service enthusiastically. Trip frequency exceeded projections. The data suggested that each shared car2go vehicle replaced between four and eleven privately owned vehicles, a finding that became the centerpiece of nearly every subsequent investor presentation and municipal negotiation. By early 2009, Daimler expanded the pilot to Austin, Texas — the first North American market — and then to Hamburg, Germany, establishing the template for a city-by-city rollout that would accelerate over the next five years.
Car2go's geographic footprint, 2008–2019
2008Pilot launches with 50 Smart cars in Ulm, Germany. 200 test users.
2009Expands to Austin, Texas — first North American market. Hamburg follows.
2011Enters Washington D.C., San Diego, and multiple German cities. Fleet exceeds 3,000 vehicles.
2013Reaches 500,000 registered members. Present in 18 cities across Europe and North America.
2014Launches in five additional cities. Begins upgrading fleet from Smart ForTwo to include Mercedes CLA and GLA models.
2016Claims 2.2 million members, 14,000 vehicles. Peak geographic footprint of 26 cities.
2018Daimler announces merger of car2go with BMW's DriveNow to form SHARE NOW joint venture.
The Density Paradox
Every city car2go entered presented the same fundamental equation, and in almost every city, the equation refused to balance. The mathematics were brutal in their simplicity: a free-floating carsharing service requires a minimum fleet density — vehicles per square mile of service area — sufficient to ensure that a potential user can find a car within walking distance at the moment of need. Too few cars, and users open the app, see nothing nearby, close the app, and call an Uber. Too many cars, and utilization drops below the threshold at which per-minute revenue covers per-vehicle costs. The sweet spot was vanishingly narrow, and it shifted constantly based on time of day, weather, neighborhood, and the competitive dynamics of whichever ride-hailing service was currently burning venture capital to subsidize fares in the same market.
In compact European cities — Berlin, Hamburg, Munich, Vienna — the math worked better, if not well. Dense urban cores with expensive parking, strong public transit systems that carsharing could complement rather than compete with, and cultural familiarity with small cars created conditions where utilization rates could approach 20–25%. A Smart ForTwo rented at €0.29 per minute in a city where it completed six to eight trips per day, averaging 18 minutes each, could generate €30–€40 in daily revenue against daily costs (depreciation, insurance, parking permits, cleaning, maintenance, fuel, telematics) that industry analysts estimated at €25–€35. Thin margins. Theoretically viable at scale.
North American cities were a different organism entirely. Austin, Denver, Portland, Washington D.C. — these were sprawling metropolitan areas where distances between destinations measured in miles, not blocks. Service areas had to be enormous to be useful, which meant fleet density was inherently low unless Daimler was willing to deploy thousands of vehicles per city. Parking was less expensive, which reduced one of carsharing's key value propositions. And the competitive landscape included not just Uber and Lyft — who were subsidizing rides at below cost to build network effects — but also the American consumer's deeply embedded preference for private vehicle ownership, a cultural attachment that proved far more durable than any McKinsey slide had predicted.
The utilization data told the story. In North American markets, car2go vehicles sat idle for an estimated 80–85% of available hours. Each vehicle might complete three to four trips per day. At $0.41 per minute and an average trip duration of 15 minutes, that generated roughly $18–$25 in daily revenue per vehicle. Against estimated daily per-vehicle costs of $30–$45 — higher than European markets due to American insurance premiums, larger service areas requiring more relocation logistics, and the higher cost of municipal operating permits — every vehicle in the North American fleet was, on average, losing money every single day it sat on the street.
The math never worked in most American cities. The question was always how long Daimler would subsidize the experiment.
The Platform That Wasn't
What distinguished car2go from the technology companies it envied — Uber, Lyft, Airbnb — was a structural fact that no amount of app design or marketing could overcome: car2go owned its assets. This was not an incidental feature of the business model. It was the business model's defining constraint.
Uber owned no cars. Airbnb owned no apartments. Their genius — and their margin structures — derived from the insight that you could build a coordination platform on top of other people's capital, capturing a percentage of each transaction while externalizing the costs of depreciation, maintenance, insurance, and idle time to the asset owners. The platform's variable cost of serving an incremental user was effectively zero. Uber's marginal cost of one more ride was the driver's time and gasoline, neither of which Uber paid for.
Car2go had no such luxury. Every vehicle in its fleet was a Daimler-owned asset depreciating on a Daimler balance sheet. When a car sat unused on a Tuesday afternoon in Denver — which was most Tuesday afternoons in Denver — the meter was running in the wrong direction. Depreciation didn't pause. Insurance premiums didn't adjust in real time to utilization. The parking permit the city of Denver charged for each vehicle to occupy public street space applied whether the car moved or not.
This is the fundamental tension that Oliver Gassmann, Karolin Frankenberger, and Michaela Csik explore in
The Business Model Navigator — the idea that business model innovation often requires recombining familiar patterns in unfamiliar ways, and that the most dangerous failure mode is adopting the language of a new pattern while retaining the cost structure of the old one. Car2go spoke the language of platform-era mobility — "access over ownership," "mobility as a service," "the sharing economy" — but its cost structure was that of a fleet operator. It was a rental car company with better UX and worse utilization.
Daimler recognized this tension, at least partially. In 2016, the company began experimenting with dynamic pricing — surge-style multipliers during peak demand — and introduced monthly membership packages that offered lower per-minute rates in exchange for recurring revenue. Neither intervention fundamentally altered the unit economics. Dynamic pricing improved revenue per trip during high-demand periods but did nothing to reduce the cost of idle vehicles during the 80% of hours when demand was insufficient. Membership packages increased customer stickiness but attracted the highest-frequency users — precisely the customers whose heavy usage drove maintenance costs up and effective per-minute revenue down.
The Competitor That Couldn't Be Out-Subsidized
The timing of car2go's North American expansion — 2011 through 2016 — placed it in direct competition with the most aggressively funded startups in the history of venture capital. Uber raised $24.7 billion in private funding before its 2019 IPO. Lyft raised $5.1 billion. Both companies were engaged in a price war that made urban transportation artificially cheap for consumers, creating a benchmark against which carsharing's per-minute pricing looked expensive and its user experience — walk to a car, drive yourself, find parking — looked burdensome.
For a 20-minute crosstown trip in Austin in 2016, a car2go user would pay approximately $8.20 ($0.41 × 20 minutes), plus the inconvenience of walking to the nearest available vehicle (which might be six blocks away), driving in unfamiliar traffic, and finding a legal parking spot at the destination. An UberX ride for the same trip cost $7–$10 — comparable in price, but the car came to you, someone else drove, and you were dropped at your exact destination. No parking. No walking. No cognitive load.
Car2go was competing not just on price but on a fundamentally less convenient proposition, and it was competing against companies with functionally unlimited capital willing to lose billions annually to acquire users. Daimler, for all its resources — €167 billion in 2018 revenue, the backing of one of Germany's most powerful industrial enterprises — was not structured to sustain indefinite losses in a non-core business unit. Quarterly earnings calls with institutional shareholders demanding profitability from a luxury automaker created a fundamentally different incentive structure than the venture-backed growth mandates that governed Uber and Lyft.
The BMW Marriage
By 2018, the strategic calculus had shifted enough that Daimler took an extraordinary step: it merged car2go with DriveNow, BMW's competing free-floating carsharing service, to form a joint venture called SHARE NOW. Two of the world's fiercest automotive competitors — companies whose brand identities were built on differentiation, whose entire market positioning depended on the consumer perceiving a meaningful distinction between a Mercedes-Benz and a BMW — agreed to combine their mobility services into a single entity.
The deal, announced in March 2018 and finalized in February 2019, created a combined fleet of approximately 20,000 vehicles across 31 cities, with over 4 million registered users. It also created a broader mobility services portfolio alongside SHARE NOW: REACH NOW (multimodal planning), FREE NOW (ride-hailing, built on Daimler's acquisition of MyTaxi), CHARGE NOW (EV charging), and PARK NOW (digital parking). The aspiration was nothing less than a comprehensive mobility super-app that could challenge Uber's dominance through the combined brand power and capital of two automotive giants.
Both companies are investing more than €1 billion to develop and expand these services, creating an intelligent ecosystem of connected mobility services for their customers.
The joint venture lasted barely two years in its original ambition. SHARE NOW began closing cities almost immediately — first pulling out of North America entirely in December 2019, then exiting London, Brussels, and Florence in early 2020. By mid-2020, the combined service operated in 16 European cities, down from 31 at launch. COVID-19 accelerated the retreat, but the structural problems preceded the pandemic. The merger had combined two money-losing operations into one larger money-losing operation. Synergies — shared technology platforms, combined fleet management, reduced per-city overhead — existed on spreadsheets but proved difficult to realize when two corporate cultures, two vehicle brands, and two legacy technology stacks had to be integrated while simultaneously managing fleet operations across multiple countries.
In 2022, Stellantis acquired a majority stake in SHARE NOW, effectively ending the Daimler-BMW experiment in shared ownership of shared mobility. The car2go brand, which had been subsumed into SHARE NOW in 2019, was already gone — a ghost fleet, dissolved into the broader narrative of automotive companies grappling with a future they could describe but not profitably operate.
The Municipal Chess Game
One dimension of car2go's story that receives insufficient attention is the regulatory architecture it had to navigate — or, more precisely, construct from scratch. When car2go launched in 2008, no regulatory framework existed for free-floating carsharing. The concept didn't fit neatly into any existing category. It wasn't a taxi service. It wasn't a traditional car rental (which operated from fixed locations with contracts and insurance waivers). It wasn't public transit. It was something genuinely new, and every city car2go entered required bespoke negotiations with municipal authorities over parking permits, service area boundaries, fleet size caps, data sharing requirements, and — critically — the use of public street space for private commercial purposes.
In some cities, car2go secured remarkable concessions. In Washington D.C., the District Department of Transportation allowed car2go vehicles to park at any legal metered space without paying the meter — a subsidy worth millions annually in foregone parking revenue that the city justified on the basis of car2go's environmental claims (fewer private vehicles, reduced emissions, decreased demand for parking infrastructure). In Austin, the city waived certain parking restrictions and granted operating permits that gave car2go preferential access to downtown spaces.
These regulatory wins were genuine competitive moats — but they were moats that required constant maintenance. Every city council election, every new transportation director, every shift in political wind created the potential for permit revocations or fee increases. And as ride-hailing companies began generating their own municipal controversies — congestion, labor disputes, safety concerns — the regulatory environment for all new mobility services grew more skeptical. By 2018, several cities were reconsidering the terms of their carsharing agreements, asking harder questions about the actual versus claimed environmental benefits, and demanding more granular data about fleet utilization, parking patterns, and trip demographics.
The Data That Daimler Really Wanted
There is a reading of car2go's history that reframes the entire venture not as a failed mobility business but as an extraordinarily expensive market research operation. From 2008 to 2019, car2go collected granular data on millions of urban transportation decisions: where people went, when they went, how long they stayed, which routes they chose, how they responded to pricing changes, what weather conditions suppressed or stimulated demand. This data — trip origins and destinations, temporal patterns, geographic heat maps of demand — was precisely the information an automaker would need to design future products, plan autonomous vehicle deployment, and understand the behavioral economics of urban mobility.
Daimler's internal research publications during this period frequently referenced car2go data as an input to autonomous driving strategy. The theory was simple: if you knew where shared vehicles were most needed, at what times, and for what trip types, you could design an autonomous fleet deployment strategy that eliminated the most expensive component of the cost structure — the vehicles' idle time — by routing self-driving cars to demand in real time rather than waiting for users to find stationary vehicles.
Whether this data justified the estimated €500 million investment is a question that depends entirely on how valuable you believe autonomous vehicle deployment intelligence to be and on what timeline you expect autonomous vehicles to actually materialize. By 2024, that timeline had stretched considerably further than anyone projected in 2016.
What the Sharing Economy Got Wrong About Cars
The deeper lesson of car2go is not about Daimler's execution, which was competent if unimaginative, or about the competitive dynamics of ride-hailing, which were historically anomalous. The deeper lesson is about the nature of automobiles as shared assets and the specific ways in which cars differ from the other categories — apartments, office space, computing power, bicycles — that have been successfully shared.
A car is the second most expensive purchase most consumers make. It is also, unlike a home, a rapidly depreciating asset that requires continuous maintenance, regulatory compliance (insurance, registration, inspection), and dedicated storage space. When you share a car between multiple users, you do not eliminate these costs — you redistribute them. And the redistribution introduces new costs that private ownership avoids: the technology platform to coordinate access, the cleaning and maintenance required between users, the insurance premiums that reflect the higher risk profile of multiple drivers, the repositioning logistics to ensure fleet distribution matches demand patterns.
The sharing economy thesis assumed that utilization was the key variable — that a privately owned car sitting idle 95% of the time represented an enormous pool of latent value that could be unlocked through coordination. This was true in the abstract and false in the specific. The 95% idle time of a privately owned car was not waste in the conventional sense. It was optionality — the knowledge that the car was available whenever the owner needed it, wherever they had parked it, without coordination costs. Carsharing replaced this optionality with a probabilistic model: the car might be nearby when you need it. For routine, predictable, low-stakes trips — the kind that made up the bulk of car2go usage — this was often sufficient. For the trips that actually determined whether a household could give up its second car — the emergency pediatrician visit at 2 a.m., the airport run with four suitcases, the Costco trip that required a trunk — carsharing was structurally inadequate.
Carsharing works best as a complement to car ownership and public transit, not as a substitute. The households that reduce their vehicle count are typically those that already had access to robust alternatives.
The Balance Sheet as Epitaph
Daimler never disclosed car2go's standalone financials. The operation was buried within the company's "Daimler Mobility" segment, later renamed "Daimler Financial Services," a division that also encompassed vehicle leasing, fleet management, and financial services — businesses profitable enough to obscure the carsharing unit's losses. Industry analysts estimated car2go generated annual revenue of €150–€200 million at its peak against operating costs that exceeded revenue by 30–50%, implying annual operating losses of €50–€100 million. Over eleven years of operation, cumulative losses likely exceeded €500 million, depending on how capital costs and corporate overhead allocations are treated.
For Daimler — a company with €172.7 billion in revenue in 2019 — €500 million spread over a decade was not an existential cost. It was, as one analyst put it, "a rounding error with a marketing budget." And car2go did deliver marketing value: it kept Daimler in the conversation about urban mobility, generated favorable press coverage, and provided a tangible response to the recurring analyst question about how legacy automakers would survive the transition to mobility-as-a-service. But as a standalone business, car2go never approached profitability in any market except possibly a handful of dense European cities, and even those were marginal.
The final chapter was quiet. In February 2022, Stellantis — the auto conglomerate formed from the merger of Fiat Chrysler and PSA Group — acquired a controlling stake in SHARE NOW. Daimler (by then renamed Mercedes-Benz Group AG) and BMW retained minority positions but effectively stepped back from operational involvement. The carsharing fleet, reduced from its peak of 20,000 vehicles to roughly 10,000, continued to operate in a dozen-odd European cities under the SHARE NOW brand. The car2go name existed only in Wikipedia entries and transportation research papers.
On the streets of Ulm, where it all began, Smart ForTwos still appeared occasionally — but they were privately owned, parked in driveways, driven by individuals who had purchased them the old-fashioned way. The 50 original car2go vehicles from that 2008 pilot had long since been decommissioned, their telematics units stripped, their data absorbed into servers in Stuttgart. The experiment was over. The data remained.
Car2go's trajectory — from visionary pilot to continental retreat — offers an unusually clean set of operating lessons precisely because the failure was structural rather than accidental. The principles below are extracted from what went right, what went wrong, and what operators in adjacent spaces can learn from a company that saw the future correctly but couldn't make the present pay for it.
Table of Contents
- 1.Don't confuse owning the asset with owning the customer.
- 2.Density is the product.
- 3.Pilot in the lab, not in the wild.
- 4.Match the cost structure to the business model's generation.
- 5.Regulatory moats are real — and fragile.
- 6.Kill optionality to create commitment.
- 7.Don't merge two losing businesses and call it synergy.
- 8.The hardest competition is a subsidized substitute.
- 9.Treat the sunk cost as intelligence, not regret.
- 10.When the thesis requires autonomy, price in the delay.
Principle 1
Don't confuse owning the asset with owning the customer.
Car2go's fatal structural flaw was that Daimler, a company whose entire history was built on manufacturing and selling physical assets, attempted to build a platform business while retaining the asset-heavy cost structure of its core business. The instinct was understandable — Daimler made cars, so of course the carsharing fleet would be Daimler cars. This gave the company quality control, brand consistency, and the ability to source vehicles at cost. It also gave the company a balance sheet loaded with depreciating metal.
The platform businesses that dominated the 2010s — Uber, Airbnb, Amazon Marketplace — succeeded precisely because they separated the coordination layer from the asset layer. They owned the customer relationship and the matching algorithm. Someone else owned the car, the apartment, the inventory. This asset-light model wasn't merely a financial preference; it was the structural prerequisite for the kind of rapid scaling and margin expansion that venture-backed growth demanded.
Car2go could have explored alternative models: partnering with independent fleet owners, enabling peer-to-peer carsharing on Daimler vehicles, or structuring the fleet as a financial product (leases to a special-purpose vehicle) rather than an operating asset. None of these approaches would have been easy, and all would have introduced new risks. But they would have aligned the cost structure with the business model's logic rather than its parent company's manufacturing identity.
Benefit: Asset-light models scale faster, fail cheaper, and align incentives between the platform and its suppliers by distributing risk.
Tradeoff: Ceding asset ownership means ceding quality control, brand consistency, and the ability to guarantee supply. Uber's driver shortages and Airbnb's quality variance are direct consequences of asset-light architecture.
Tactic for operators: Before building anything, draw two org charts — one where you own the core asset, one where you coordinate other people's assets. If the unit economics only work in the second chart, don't build the first one because it feels safer.
Principle 2
Density is the product.
In free-floating carsharing, the user experience is almost entirely determined by one variable: how far they have to walk to the nearest available vehicle. Car2go's internal data showed that demand collapsed when walking distance exceeded five minutes. Below five minutes, usage scaled reliably. This created an iron requirement for fleet density — vehicles per square mile of service area — that was expensive to achieve in sprawling North American cities and manageable only in compact European cores.
The implication extends far beyond carsharing. Any marketplace or distributed-asset business faces a version of the density problem: the product's value to each user depends on how many other participants (or assets) are nearby. Ride-hailing has the same dynamic (driver ETA determines conversion). Food delivery has it (restaurant selection within delivery radius). Even SaaS products exhibit a version of it when network effects are localized (
Slack is useless in a company of one).
How fleet density correlated with user retention in car2go markets
| Market Type | Avg. Walking Distance | Utilization Rate | Retention (6-mo.) |
|---|
| Dense European (Berlin, Vienna) | ~3 min | 20–25% | ~55% |
| Mid-density (Hamburg, Portland) | ~6 min | 15–18% | ~35% |
| Low-density U.S. (Denver, Austin) | ~10 min | 10–15% | ~20% |
Benefit: Solving for density first creates a virtuous cycle — more available vehicles attract more users, whose trip patterns distribute vehicles more effectively, reducing the need for manual repositioning.
Tradeoff: Density requires concentrated capital deployment. You cannot be everywhere at once. Car2go's mistake was expanding to 26 cities with insufficient density in most of them, rather than achieving dominant density in five.
Tactic for operators: Launch in the smallest viable geography and saturate it before expanding. Define a single quantitative density threshold (vehicles per square mile, drivers per zip code, restaurants per delivery zone) and refuse to expand beyond the boundary until you've hit it within it.
Principle 3
Pilot in the lab, not in the wild.
The Ulm pilot — 50 cars, 200 users, a city of 120,000 in Daimler's backyard — produced data that was accurate for Ulm and misleading for everywhere else. The small-city pilot confirmed demand but couldn't test the variables that would determine success or failure in target markets: insurance costs in American cities, competition with ride-hailing, municipal parking politics, fleet management logistics at scale, or consumer behavior in sprawling metropolitan areas.
Car2go scaled from a pilot that proved the concept to full-market launches that tested entirely different hypotheses. The 2009 Austin launch wasn't a scaled version of Ulm. It was a different experiment in a different regulatory, competitive, and geographic context, but it was treated as a rollout rather than a test.
Benefit: Rigorous pilot design — where the test market closely resembles the target market in every structural dimension — prevents the most expensive form of confirmation bias: scaling a hypothesis that was validated in the wrong conditions.
Tradeoff: Finding a true analog market is hard, and waiting for perfect pilot conditions can cost first-mover advantage.
Tactic for operators: For every pilot, list the five assumptions that must be true for the business to work at scale. Design the pilot to stress-test the most dangerous assumption — the one whose failure would kill the business — not the one most likely to confirm your thesis.
Principle 4
Match the cost structure to the business model's generation.
Gassmann, Frankenberger, and Csik argue in
The Business Model Navigator that business model innovation typically involves recombining existing patterns — and that 90% of all business model innovations are fundamentally recombinations of 55 identifiable patterns. Car2go combined elements of "Pay Per Use" (per-minute pricing), "Flatrate" (in some membership tiers), and "Fractionalized Ownership" (shared access to an asset). The patterns were right. The cost structure was inherited from the parent company rather than designed for the new model.
A business model pattern is only as effective as the cost structure it sits atop. Car2go adopted the usage model of a software platform while retaining the cost structure of a physical fleet operator. The result was a business that looked like a tech startup in its customer-facing layer and behaved like a capital-intensive industrial operation in its P&L.
Benefit: Deliberately designing cost structure to match the business model's logic — rather than inheriting it from the parent organization — prevents the most common failure mode of corporate innovation: new business models strangled by old cost assumptions.
Tradeoff: Designing a novel cost structure often requires organizational separation, which reduces the synergies (procurement, brand, distribution) that justified the corporate parent's involvement in the first place.
Tactic for operators: Before launching a new business model, calculate the fully loaded cost per unit of revenue. If that cost per unit is more than 2× what a purpose-built startup would achieve, you have a structural problem that no amount of optimization will fix. Redesign the model or don't launch.
Principle 5
Regulatory moats are real — and fragile.
Car2go's ability to negotiate preferential parking access in cities like Washington D.C. and Austin constituted a genuine competitive advantage — one that later entrants would have struggled to replicate. These regulatory relationships represented years of lobbying, data sharing, and trust-building with municipal authorities. In some markets, car2go's first-mover regulatory status was its most defensible asset.
But regulatory moats are uniquely vulnerable to political cycles, public sentiment shifts, and the regulatory contagion effect — where controversy in one city (ride-hailing safety incidents, gig economy labor disputes) poisons the regulatory environment for all mobility services in other cities. Car2go's parking privileges in D.C. lasted precisely as long as the political consensus that carsharing deserved preferential treatment for environmental reasons. When that consensus weakened, the moat drained.
Benefit: Regulatory advantages compound over time and are extremely difficult for competitors to replicate quickly. A company with established municipal relationships can expand into adjacent regulatory negotiations using credibility from existing permits.
Tradeoff: You are building on someone else's land. Political winds shift. The regulatory advantage that took three years to build can be revoked in a single council session.
Tactic for operators: Treat regulatory wins as a wasting asset. From the day you secure a permit, begin building a second layer of defensibility — brand loyalty, network effects, data advantages — that will sustain the business if the regulatory moat drains.
Principle 6
Kill optionality to create commitment.
Car2go's biggest competitor was not Uber, Lyft, or public transit. It was the second car in the household garage. The core economic thesis of free-floating carsharing — that it was cheaper than owning a vehicle — was only true if the household actually sold the vehicle. Most car2go users treated the service as a complement to existing transportation options rather than a substitute. They used car2go for the occasional short trip and kept their car for everything else. This meant car2go captured the lowest-value trips (short, spontaneous, discretionary) while the highest-value use cases (commuting, family transportation, weekend trips) remained with the private vehicle.
The service needed to be so reliable, so convenient, and so comprehensive that users would feel comfortable selling a car. It never got there. The probabilistic availability model — the car might be near you — couldn't compete with the deterministic availability of a vehicle parked in your driveway.
Benefit: Products that create genuine commitment from users — by making it rational to abandon alternatives — capture dramatically more lifetime value than products that coexist politely with the status quo.
Tradeoff: The threshold for commitment-level reliability is far higher than the threshold for trial adoption. Building for commitment is 10× more expensive per user than building for trial.
Tactic for operators: Identify the one behavior change that would make your product indispensable rather than supplementary. Then work backward from that behavior change to determine the minimum product reliability required to trigger it. If you can't fund that level of reliability, you're building a nice-to-have.
Principle 7
Don't merge two losing businesses and call it synergy.
The SHARE NOW joint venture between Daimler and BMW combined two money-losing carsharing operations and promised that scale efficiencies would close the gap. They didn't. The merger added organizational complexity — two corporate cultures, two technology platforms, two vehicle brands with different maintenance requirements — without fundamentally changing the unit economics of any individual market. A car2go vehicle and a DriveNow vehicle in the same city were now managed by the same entity, but they still had the same depreciation rate, the same insurance costs, and the same utilization problem.
Mergers between sub-scale, unprofitable competitors can work when the underlying unit economics are sound and the primary constraint is overhead cost or competitive spending. They fail when the underlying unit economics are broken and the merger simply creates a larger entity losing money at larger scale.
Benefit: Merging genuinely complementary operations — where one party has distribution and the other has product — can create value that neither could achieve alone.
Tradeoff: Mergers between operationally similar, unprofitable businesses primarily produce accounting consolidation, not operational improvement. The integration costs consume whatever overhead savings the combination generates.
Tactic for operators: Before any partnership or merger, calculate the pro forma unit economics of the combined entity at the market level. If the combined unit economics are still negative, the merger is a delay tactic, not a strategy.
Principle 8
The hardest competition is a subsidized substitute.
Car2go entered North American markets just as Uber and Lyft began their venture-subsidized assault on urban transportation. The ride-hailing companies were not just competing on price; they were competing on convenience, reliability, and the elimination of driving and parking — tasks that car2go required users to perform themselves. And they were doing so while losing billions of dollars annually, funded by investors who believed that market share acquired today would generate profits tomorrow.
Competing against a subsidized substitute is fundamentally different from competing against a profitable competitor. A profitable competitor has a floor — they won't price below cost indefinitely. A subsidized competitor has no floor except their investors' patience, which in the case of Uber turned out to extend for more than a decade and more than $24 billion.
Benefit: Understanding whether your competitor is profitable or subsidized changes every strategic calculation — pricing strategy, market entry timing, capitalization requirements, and expected time to profitability.
Tradeoff: You cannot simply wait for the subsidized competitor to die. They may reach profitability before they run out of capital, or they may permanently reset consumer expectations about price and convenience.
Tactic for operators: Map your competitive landscape into two columns: competitors that must be profitable to survive and competitors that can lose money indefinitely. Your strategy against each is fundamentally different. Against the first, compete on product. Against the second, compete on defensibility and patience.
Principle 9
Treat the sunk cost as intelligence, not regret.
Daimler collected eleven years of granular urban transportation data through car2go — trip patterns, demand curves, pricing elasticities, behavioral responses to weather, events, and time of day across 26 cities. This data, while insufficient to justify the investment on a standalone basis, represented a genuine strategic asset for Daimler's autonomous vehicle and future mobility planning.
The lesson is that failed business ventures in adjacent spaces can generate intelligence — about customers, markets, operations, and technology — that has value beyond the venture's own P&L. The key is structuring the data capture intentionally from day one, not as an afterthought.
Benefit: Intentional intelligence harvesting from adjacent ventures creates optionality that may be worth multiples of the venture's operating losses, particularly when the intelligence informs a much larger core business.
Tradeoff: "This is really a data play" is the most common rationalization for a failing business. The intelligence justification must be specific and quantifiable, not a retrospective consolation prize.
Tactic for operators: For any exploratory venture, define three to five specific intelligence outputs that would be valuable to the core business regardless of the venture's financial performance. Instrument the venture to capture this intelligence from day one. If the venture fails, you have assets. If it succeeds, you have both.
Principle 10
When the thesis requires autonomy, price in the delay.
Much of car2go's long-term strategic justification rested on the assumption that autonomous vehicles would eventually transform the economics of carsharing. An autonomous free-floating fleet could reposition itself to match demand in real time, eliminate the insurance costs associated with human drivers, and achieve utilization rates of 50% or higher. This was the endgame that made the near-term losses tolerable — in theory.
The autonomous vehicle timeline has slipped repeatedly. In 2016, Daimler projected Level 4 autonomous vehicles by the early 2020s. As of 2024, Level 4 autonomy remains limited to geofenced test deployments in a handful of cities. Every year of delay extended the period of operating losses that the carsharing business had to endure.
Benefit: Building infrastructure (data, regulatory relationships, brand recognition, user habits) in advance of a technology shift can create enormous first-mover advantage if the timing is right.
Tradeoff: If the technology shift takes twice as long as expected — which in the case of autonomous vehicles, it has — the infrastructure costs accumulate without the transformative economic improvement that justifies them.
Tactic for operators: For any strategy that depends on an anticipated technology shift, build a financial model with three timelines: the optimistic case, the base case, and a "double the delay" case. If the business doesn't survive the third scenario, you need a bridge strategy that generates returns independent of the technology shift. Hope is not a timeline.
Conclusion
The Expensive Education of an Industrial Giant
Car2go's ten principles coalesce around a single insight: that business model innovation within an established corporation is not a product development challenge but a structural redesign challenge. The product — free-floating carsharing — was sound. Consumer demand was real. The technology worked. What didn't work was the attempt to operate a platform-era service on an industrial-era cost structure, within an industrial-era governance framework, against competitors whose entire organizational design was optimized for the new model.
The operators who extract the most value from car2go's experience will be those building asset-intensive businesses in the era of asset-light competition. The principles above are not prescriptions for success — they are diagnostic tools for identifying structural mismatches between ambition and architecture before those mismatches consume hundreds of millions in capital.
Daimler learned something from car2go. The question is whether the tuition was worth it.
Part IIIBusiness Breakdown
The Business at a Glance
Final State
Car2go / SHARE NOW — Legacy Metrics
~10,000Vehicles in fleet (SHARE NOW, post-consolidation, 2022)
~12Operating cities (European markets only)
~3.4MRegistered members (SHARE NOW combined)
€150–200MEstimated peak annual revenue (car2go standalone)
NegativeOperating profitability (never achieved at group level)
StellantisMajority owner following 2022 acquisition
Car2go, as a standalone entity, no longer exists. Its operations were absorbed into SHARE NOW in 2019, which itself was partially divested by Daimler and BMW in 2022. For analytical purposes, the business breakdown below reconstructs the economics of car2go at its operational peak (2016–2018) and examines the SHARE NOW successor as of 2022–2023 to provide a current state assessment. The financial data relies on analyst estimates and industry reports, as neither Daimler nor BMW disclosed car2go/SHARE NOW's standalone financials.
The business, at its peak, represented one of the largest experiments in urban shared mobility ever attempted by a legacy automaker. Its approximately 14,000 vehicles across 26 cities served over 3 million registered members, generating an estimated €150–€200 million in annual revenue. It never achieved group-level operating profitability.
How Car2go Made Money
Car2go's revenue model was deceptively simple: users paid for vehicle access on a per-minute basis, with daily and hourly caps that functioned as de facto rate limiters. Revenue streams, such as they were, fell into three categories:
Car2go revenue streams at peak operations (2016–2018 estimated)
| Revenue Stream | Estimated Annual Revenue | % of Total | Trend |
|---|
| Per-minute usage fees | €120–160M | ~80% | Declining per-trip |
| Membership / package plans | €15–25M | ~10–12% | Stable |
| Late fees, damage charges, other | €10–20M | ~8–10% | Stable |
Per-minute usage was the core revenue driver. Rates varied by market and vehicle class — €0.24–€0.31 per minute for Smart ForTwo vehicles in European markets, $0.35–$0.41 per minute in U.S. markets, and higher rates for Mercedes-Benz CLA/GLA premium vehicles introduced in 2016. Average trip duration was 15–20 minutes, generating €4–€8 per trip in Europe and $5–$8 per trip in North America.
Membership packages, introduced in 2016, offered discounted per-minute rates (typically 20–30% below standard pricing) in exchange for monthly fees of €9–€19. These packages improved revenue predictability but concentrated usage among high-frequency riders whose trips generated the lowest marginal revenue per minute.
Unit economics at the vehicle level were the central challenge. A typical car2go vehicle in a North American market generated an estimated $18–$25 in daily revenue against $30–$45 in daily fully loaded costs (depreciation, insurance, parking permits, maintenance, cleaning, fuel, telematics, corporate overhead allocation). In the best-performing European markets, the ratio was closer to balanced — €30–€40 in daily revenue against €25–€35 in daily costs — but even these markets operated at thin or negative margins after corporate overhead.
Competitive Position and Moat
Car2go operated in a competitive landscape that shifted dramatically during its eleven-year life. The relevant competitor set included:
Key competitors by category, 2016–2019
| Competitor | Model | Fleet/Scale | Funding |
|---|
| Uber | Ride-hailing (asset-light) | 3.9M drivers (2018) | $24.7B private |
| Lyft | Ride-hailing (asset-light) | 1.9M drivers (2018) | $5.1B private |
| Zipcar (Avis) | Station-based carsharing | ~12,000 vehicles | Avis Budget Group |
| DriveNow (BMW) | Free-floating carsharing | ~6,000 vehicles |
Car2go's competitive moats, where they existed, were:
- Brand recognition in carsharing. Car2go was the first major free-floating carsharing brand globally and retained significant brand awareness in European markets through 2019. Recognition did not translate into pricing power or switching costs.
- Municipal relationships and parking permits. The regulatory agreements car2go negotiated in each city represented real barriers to entry — competing services had to negotiate their own terms from scratch. But these agreements were revocable and non-transferable.
- Scale in key European markets. In Berlin, Hamburg, Munich, and Vienna, car2go/SHARE NOW achieved sufficient fleet density to create a reasonably reliable user experience. These remained the only markets where utilization approached economic viability.
- Daimler vehicle supply at cost. Access to Daimler-manufactured vehicles at internal transfer prices reduced fleet acquisition costs by an estimated 15–20% relative to market pricing. This advantage was real but insufficient to offset the fundamental utilization problem.
Where the moat was weak: car2go had no network effects (more users didn't improve the service for other users beyond marginal fleet redistribution), no meaningful switching costs (users could and did maintain accounts with multiple mobility services simultaneously), and no data advantage that translated into a competitive product improvement cycle.
The Flywheel
Car2go's intended flywheel — the reinforcing cycle that was supposed to compound its advantages over time — was theoretically elegant and practically stalled.
The reinforcing cycle car2go needed — and largely failed — to achieve
Step 1Deploy dense fleet in compact urban areas → users can reliably find vehicles within 5-minute walk.
Step 2Reliable availability drives user adoption and frequency → registered members grow, trips per vehicle increase.
Step 3Higher utilization improves per-vehicle economics → unit costs decline, enabling lower pricing or expanded fleet.
Step 4Lower prices / larger fleet attract more users → further utilization improvement.
Step 5User trip data improves fleet positioning algorithms → vehicles are repositioned to anticipate demand, reducing idle time.
Step 6Improved economics fund expansion to adjacent neighborhoods and cities → geographic density compounds.
The flywheel stalled primarily at the transition from Step 2 to Step 3. User adoption grew — 3 million registered members was a significant base — but utilization rates never reached the threshold at which unit economics turned positive in most markets. Without positive unit economics, pricing could not be lowered, fleet could not be expanded, and the cycle could not accelerate. The flywheel became a treadmill: more users generated more trips, but each trip consumed more in operating costs than it produced in revenue.
In the strongest European markets (Berlin, Hamburg), the flywheel achieved partial rotation — utilization approached 20–25%, pricing was sustainable, and fleet positioning algorithms improved over time. But even in these markets, profitability was marginal, and the flywheel never reached the escape velocity that would have allowed car2go to fund its own expansion from operating cash flow.
Growth Drivers and Strategic Outlook
For SHARE NOW, the successor entity, the remaining growth vectors are constrained but identifiable:
-
Electrification of the fleet. SHARE NOW has been transitioning to electric vehicles, with over 25% of its fleet battery-electric by 2023. EVs have lower per-mile operating costs (fuel savings of ~€0.05/km versus ICE vehicles), which improves unit economics — but higher upfront acquisition costs and charging infrastructure requirements partially offset this advantage. The TAM for urban EV carsharing in Europe is estimated at €2–€4 billion by 2030.
-
Subscription-based models. The shift from pure per-minute pricing to subscription tiers (daily, weekly, monthly all-access passes) creates more predictable revenue and higher average revenue per user. SHARE NOW has expanded its subscription offerings since 2021.
-
Integration with public transit. Partnerships with municipal transit authorities — offering carsharing as a "last mile" complement to bus and rail networks — could drive incremental utilization during off-peak hours. Several European cities are piloting integrated mobility passes that include carsharing minutes alongside transit fares.
-
Corporate fleet replacement. B2B offerings — where companies replace employee vehicle fleets with carsharing memberships — represent a higher-value, more predictable customer segment than individual consumers. Corporate clients generate more consistent utilization and longer contract commitments.
-
Expansion within existing markets. Rather than entering new cities (the strategy that over-extended car2go), SHARE NOW can increase fleet density and service area within its existing 12 European markets, improving utilization without the overhead of new-market entry.
Key Risks and Debates
The risks facing SHARE NOW — and the broader free-floating carsharing model — remain structural:
-
Persistent negative unit economics in most markets. The fundamental math that defeated car2go — high fixed costs per vehicle, low utilization, pricing constrained by ride-hailing alternatives — has not been solved. SHARE NOW's consolidation to 12 cities reflects an implicit acknowledgment that the model only works in the densest European urban cores. If utilization in remaining markets deteriorates due to competitive pressure or economic downturn, there is nowhere left to consolidate.
-
Ride-hailing price normalization. As Uber and Lyft have raised prices toward profitability (Uber achieved its first annual operating profit in 2023), the pricing differential with carsharing has narrowed. But ride-hailing's convenience advantage — car comes to you, no driving required — persists. If ride-hailing achieves sustainable profitability at current price levels, carsharing's value proposition is permanently compressed.
-
Autonomous vehicle disruption. The same autonomous vehicle technology that car2go's thesis depended upon could, if deployed by a competitor (Waymo, Cruise, Tesla), destroy the remaining carsharing business model entirely. An autonomous ride-hailing fleet offers the convenience of Uber with the cost structure of carsharing — no driver cost — and eliminates the parking and walking constraints that limit carsharing adoption.
-
Stellantis ownership uncertainty. Stellantis, which acquired majority control of SHARE NOW in 2022, has not publicly committed to long-term investment in the service. The conglomerate faces its own profitability pressures and may view SHARE NOW as a non-core asset to be maintained, monetized, or eventually divested.
-
Micromobility substitution. E-scooter and e-bike services (Lime, Tier, Bolt) have captured a significant share of short urban trips — precisely the trips that constituted car2go's core use case. A 10-minute scooter ride at €0.15/minute costs less than half the equivalent car2go trip and requires no parking. In cities with good scooter infrastructure, carsharing's addressable market for short trips is shrinking.
Why Car2go Matters
Car2go matters not because it succeeded but because it failed in ways that illuminate the structural physics of platform businesses, asset-intensive service models, and corporate innovation. Three lessons endure:
First, the relationship between cost structure and business model is not a secondary consideration — it is the consideration. Car2go proved that consumer demand for a new service category can be real, the technology can work, and the brand can achieve recognition, all while the business loses money on every transaction due to a structural mismatch between how value is created (per-minute access) and how costs are incurred (per-vehicle ownership). The operators who study car2go most carefully will be those building the next generation of asset-intensive marketplace businesses — in logistics, in healthcare delivery, in energy — where the temptation to own the asset for quality-control reasons must be weighed against the economic penalty of carrying that asset on the balance sheet.
Second, competitive timing is a strategic variable as consequential as product quality or market sizing. Car2go launched into a window that seemed favorable — before ride-hailing had normalized — but scaled into a window that was catastrophic, as Uber and Lyft's venture-subsidized pricing made every alternative urban transportation service look expensive and inconvenient by comparison. The lesson is not that car2go should have launched later (the window for free-floating carsharing may never open again) but that operators must model their competitive landscape dynamically, incorporating the capital structures and incentive models of competitors, not just their current pricing and market share.
Third, the distance between a correct thesis and a profitable business can be measured in billions. Car2go's central thesis — that urban mobility is shifting from ownership to access — was and remains correct. The number of carsharing users globally continues to grow. The environmental logic is sound. The municipal interest is real. But the specific implementation — a legacy automaker's fleet of owned vehicles scattered across public streets, priced by the minute against ride-hailing apps backed by unlimited venture capital — could not bridge the gap between vision and viability.
In Stuttgart, Mercedes-Benz Group AG continues to invest in autonomous driving technology, digital services, and future mobility platforms. The lessons of car2go live somewhere in those investment memos — an expensive, granular education in the difference between knowing where the world is going and getting there profitably. The €500 million spent over eleven years bought Daimler many things: data, brand narrative, regulatory relationships, and the organizational memory of a failure disciplined enough to teach. Whether those assets compound into something valuable depends on decisions being made now, in boardrooms where the ghost fleet of car2go is invoked not as cautionary tale but as tuition paid. The meter, as always, is running.
