It's a rainy night. A man stands on the side of the road, pushes a button on his phone, an app pops open, he enters his location, and, voilà, the app tells him a car is coming to pick him up and shows him how far away it is. No searching for the local cab company's phone number and waiting on hold, no stepping out into traffic to wave around, no need for cash. It's 2008. Uber's ridesharing service doesn't exist. The app making this promise? Taxi Magic. If you don't recognize the name, that's because Uber came along and ate its lunch.

"What Uber did is incredible," says George Arison, one of the founders of Taxi Magic, as we ride in the back of an Uber from San Francisco's Castro district to two warehouses near the airport that belong to his new company, an online used-car marketplace called Shift. "They out-executed everyone. They were actually late to the game, but Travis [Kalanick] was willing to destroy his existing business with black cars to go for the peer-to-peer business. At Taxi Magic, we chose not to do that--which was a mistake."


Arison, 38, left Taxi Magic in 2010, three and a half years after it launched as a service meant for BlackBerry-using business travelers, and a year and a half after it appeared in Apple's new App Store. (The company was later purchased by Verifone and operates today under the name Curb.) He eventually cashed out his shares and made a tidy sum, but his departure came after the company essentially forfeited its leadership position in the on-demand trans­por­tation revolution by not seizing one of the biggest prizes in recent business history. Uber is now the most successful startup in the world, with a valuation of more than $60 billion.

A garrulous man with a round, perpetually stubbly face and a big grin, the self-assured Arison does not come off as humbled by the experience. He says he learned a lot from his failure to invent Uber, and that he's applying that knowledge as he builds Shift, which launched three years ago and which Arison claims is the single largest seller of used cars in the Bay Area. He's raised $74 million for Shift--from heavy hitters such as Goldman Sachs, Draper Fisher Jurvetson, and Highland Capital Partners--and the company broke even this fall in San Francisco, its largest market. (It also operates in the Los Angeles and Washington, D.C., areas.)

Certainly Arison has learned one of the hardest lessons of entrepreneurship: Take your licks and move on with hard-won wisdom. Watching how Uber succeeded is helping him avoid some of the traps that snared Taxi Magic. Shift is on track to become a good used-car business. Whether it can become a transformative force in its industry, as Arison hopes, is a different question altogether.


Shift is one of a handful of startups that are trying to reimagine the used-car market. Sales of used cars generated $390 billion in 2014, according to the latest data from Manheim, an auto-auction company. Yet the actual experience of buying or selling one tends to suck. If you sell your car to a dealer, you'll likely get thousands less than you would selling to someone on Craigslist. If you sell on Craigs­list, you'll have strangers coming to your house and uncomfortable test rides. Startups like Carvana and Vroom buy cars from people, inspect them, and then resell them at competitive rates. The margins work because the companies don't run expensive physical dealerships.

Shift's big twist is that it never owns the cars it sells. Just as Airbnb doesn't own a single room and Uber doesn't own its cars, Shift's mission is to be the world's largest seller of cars without ever owning the vast majority of its inventory. Shift also offers on-demand, at-home test drives, a first for online used-auto sales, which Arison hopes will elevate his company above the competition. If you're buying a car through Shift, an employee called a "car enthusiast" brings it to your house for a test drive, and then you simply pay the list price, no haggling. If you're selling, the Shift employee comes over to check out your car, quotes an estimated minimum price, and then takes the car away so the company's mechanics can tune it up and list it on the site. If nobody buys the car within three months, the seller gets the minimum price and Shift sells it to an auto broker. If it sells for more than the minimum, Shift splits the upside with the seller, 50-50.

As Arison sees it, Shift is a "dis­ruptive, tech-driven marketplace"--by which he means it's the Uber of used-car sales, an efficient peer-to-peer platform rather than a storefront.

Arison has always aimed high. Born in Tbilisi, Georgia, when it was still part of the Soviet Union, Arison (then named Irakly Areshidze) grew up the son of a prominent nuclear physicist and grandson of a senior Communist party official. By age 12, he says, his plan was to get a Western education and come back to be president of Georgia. He attended Middlebury College and returned to his homeland briefly to work in politics, but he'd already fallen in love with America and the opportunities it offers. He eventually settled in Washington, D.C. A friend, Toby Russell, who was working for Boston Consulting Group in New York City, helped Arison land a job with BCG in Washington. Arison began traveling to visit clients, and was often frustrated when trying to book taxis in unfamiliar cities. "There has to be a better way to do this," he thought--the classic entrepreneurial spark. "Why isn't there an application on my BlackBerry where I can book a taxi on demand?"

It was 2006. Not only had nobody thought of using smartphones for local on-demand services, but there was also really no such thing as third-party apps. The iPhone didn't exist, let alone the App Store. Arison had stumbled onto a huge idea, and he quickly brought it up to Russell. He also mentioned it to a D.C.-area entrepreneur he'd befriended, Tom Depasquale, who had recently sold a company he founded to the business-travel and expense-management company Concur. Arison, Russell, and Depasquale met one night for drinks. "You know, we manage expenses really well for air travel and hotels and meals, but we can't do it for ground travel, because it's all cash. And that's like 10 percent of all travel spending," Depasquale said. The ideas of booking a taxi and expensing a taxi fit together very nicely, he suggested.

"At some point that night, it occurred to me that we weren't just having drinks. We were founding a company," remembers Russell, who now works for Arison at Shift. "So we did. We invented the on-demand, press-a-button-and-get-a-car-and-pay-for-it-electronically space."

 inline image

Shift's operations center, called the Hub, occupies two warehouse buildings wedged between Highway 101 and the San Francisco Bay. In the garage where cars wait to be taken out for test drives, several hundred vehicles are being detailed by Shift employees, and a car-size turntable surrounded by white sheets serves as the photo studio. But perhaps the most important thing in this building is on Arison's computer. It's a document that lists everything he learned from his experience with Taxi Magic, which he's now applying to Shift.

When it started, Taxi Magic was a business-to-business service known as RideCharge. Depasquale was working full-time as the executive vice president of tech­nology at Concur, and he arranged for RideCharge to work as a plug-in to the Concur expense-reporting tool. Business travelers would use RideCharge to book and pay for their taxis, and RideCharge would add a service charge on top of the fare. Taxi companies would get the same fares as usual, travelers would have an easier time getting a ride (in theory), and their employers got cleaner expense reports.

"At some point that night, it occurred to me that we weren't just having drinks. We were founding a company."

Today, Arison sees that cozy arrangement with Concur as one of the foundational flaws of Taxi Magic--a structure built to make some cash from his idea but not to allow it to flourish. "I think you need to build a company to go public--build it with the intention of becoming really big. Maybe it ends up working, maybe not, but it's really important if you want to have a chance at attracting the best talent or the best investors. Those people want to work on things that could become massive."

Another lesson involves understanding that customers might not use your product the way you expect them to. "At Taxi Magic, we had a very strong opinion of what the product should be," Arison says. "And we went out and built it without talking enough to customers. We built all this stuff based on what we thought was right, and then we put it out onto the market and learned that a bunch of things we had built were totally wasteful."

Originally, customers had to choose to pay through the app to generate the service fee to Taxi Magic, but they just wanted to book a ride on the app. They'd pay the driver separately and then Taxi Magic would have to charge them the fee separately, which confused and annoyed customers. "We'd made this huge investment in building the payment technology, and we ended up having to throw it all out," says Arison. In the early days of Shift, on the other hand, "we had a whole operation going without having any technology built. And now we're building for what we know we need."

Depasquale says Taxi Magic's booking didn't work very well either. "The taxi companies' dispatch systems were bizarre. There could be a car across the street from a user, but that wouldn't be the one sent to pick up the fare--it would be the one that hadn't been dispatched in the longest time," he remembers. "So while that cab drove 10 blocks, the user would end up getting in another car rather than wait." Sometimes, cab drivers simply wouldn't pick up a passenger, he says, because the regular dispatch system was keeping them busy. Depasquale says more than half of the orders in those early days ended up not resulting in a ride.

When Apple introduced its App Store in 2008, the RideCharge team stopped everything and raced to change its model to go after the mass market rather than just business travelers. Rebranded as Taxi Magic, the app launched in December 2008. In one day, it got around 30,000 downloads. Apple featured it in full-page ads in The New York Times and The Wall Street Journal. "We grew quite well for a while," Arison says, "but we kept running into trouble extracting revenue from the taxi companies." In addition to tacking its service fee on to riders' bills, the new app charged taxi companies for each booking it sent them. But to the taxi companies, an online order wasn't worth more than a phone call. Their fleets were constrained by the number of taxi medallions available in a market, so as long as their drivers were busy, Taxi Magic added no value to them--even if it benefited riders.

"What the world needed was a new dispatch system," Depasquale says.

Therein lies Arison's biggest lesson by far: You can't build a disruptive technology on top of a legacy system. Arison finally decided that the solution to Taxi Magic's woes was to make an end run around the taxi-dispatch systems by creating a companion app for drivers that would effectively become a new, better dispatch system--an idea that was essentially Uber. The board, which included a taxi-fleet owner, was resistant to the idea. Taxi Magic had five co-founders--Arison, Russell, Depasquale, Keith Forsythe, and Sanders Partee--but none held the title of CEO, or the executive power to throw weight behind Arison. The board vetoed the plan.

"I bet that the industry's unwillingness to change would not be impacted by more investment" from Taxi Magic, Depasquale says now. Instead, the company refocused on building a suite of technologies for taxis, such as backseat video screens. It was mid-2010 and, having lost the power struggle, Arison left for a three-year stint at Google.

As he bided his time in Mountain View, Arison focused on the idea of vertically integrating his next company, what he calls building a "full-stack" startup, to attempt to avoid the trap Taxi Magic fell into. Arison cribs the phrase from Chris Dixon, a partner at the venture capital powerhouse Andreessen Horowitz. To fix a broken industry, Arison now believes, you have to rebuild it at every level. "It's really critical, even though it creates a more complex business," he says. More critical is ensuring that the "rebuilt" version is better than the original.

At Shift headquarters in the Castro, in a beige two-story building that's filled to capacity with young techies, Arison is leading a meeting of several executives to discuss priorities for 2017: "Fully scalable tech for ops," "captive finance," and "post-purchase" top the list. For 30 minutes, the group works to define the terms and establish concrete goals to help achieve them.

Fully scalable tech for ops, for instance, will involve refining how vehicles move from the Hub to test-drive locations and back. Currently, one employee drives one vehicle to one potential customer's house for a test drive, but it's possible the system will ultimately involve a fleet of trucks that shuffle multiple cars around, as well as software-controlled scheduling. Another big priority is captive finance. Most of the financing Shift offers to buyers comes from partner banks, but this year it rolled out its own lending system, backed by its own and some outside investor capital, which it aims to grow into a major feature. Shift currently makes most of its money on the spread between a car's guaranteed price to the seller and the list price to the buyer, but in the future it hopes additional services such as financing and warranties will be the real profit centers. The company wants to offer post-purchase services such as maintenance and repair, just as dealers do.

As Shift focuses on building out its extra services, it risks re-creating some of the flaws of the traditional dealer model that it says it wants to eliminate. Arison says customers distrust car dealers who push add-ons like upholstery treatments or undercoating. But if Shift aims to boost its margins by selling extras, what's to keep customers from distrusting Shift? Arison counters that the Shift purchase process happens on a computer instead of in a salesman's office, and that pushing unnecessary add-ons "is fundamentally against our company values."

Shift's model has one problem that pushy in-person dealers don't have. When you sell a car on Shift, the company takes the vehicle from you immediately, and until it is sold, you must continue to make insurance and loan payments. Meanwhile, you don't have your car or the proceeds from a sale to buy a new one. "There's no question there's a piece of the user base that cannot use Shift because we take the car away before you buy a new one," Arison says. "We're going to fix that"--by offering bridge loans, either through Shift's own finance arm or through partners. "But we won't be doing that today, because you can do only so many things." Arison says the issue hasn't hampered Shift's growth much. But it's an inferior experience for customers compared with the status quo. For a company that wants to reinvent a broken industry, it's not a good look.

As the full-service version of Shift takes shape, a more fundamental flaw is this: It's looking more and more like an incremental improvement in the online used-car industry, not a revolutionary, Uber-like innovation. Over a decade ago, sites like Carfax and TrueCar moved much of buyers' initial car research online. The natural next step was for new players to fuse all the search tools and information with a simple checkout process and no-haggle pricing--hence owning the customer from start to finish without having to build expensive physical dealerships. Shift takes that rebuilding of the old model online one step further than anyone else by adding on-demand, at-home test drives. It's a nice customer service improvement, but it's not clear that paying employees to ferry cars around is a better investment than brick-and-mortar stores.

Shift's innovation also makes growth harder. Though it predicted last year that it would expand its at-home test-drive model to 20 markets in 2016, Shift has added only one, the D.C. area. (It intends to add three more in 2017.) "We decided to go deep first rather than broad, so we don't find ourselves with a gun to our head because we're in 20 markets, losing money, and trying to build up all these services," says Arison. "Instead, let's be really careful with our money, get to profitability in our current markets, and then expand." (Like competitors Carvana, Vroom, and Beepi, Shift will deliver, without an at-home test drive, throughout the continental U.S. for a fee.)

Careful growth is sound business logic, but it also paints Shift's business as a chain of local used-car lots with a great website on top. There's nothing wrong with that, but it's a reshuffling of the legacy model, not a disruption. This might be the biggest Uber lesson of all, and one that Arison hasn't yet internalized: If you want to build a transformative company, you have to reimagine an industry, or create a new one, not buff up an old one. That's why Uber succeeded where Taxi Magic failed. It didn't try to improve taxis; it replaced them.

Depasquale, Arison's former mentor, has some ideas of what a truly disruptive business model might look like for Shift. "Why not think about ways to aggregate people who live near one another who might want to share a car," he says, "and maybe work on selling shared cars, where people pool money and you manage the title and the booking system for them?" In other words, reimagine the way things have always been done.

Arison points out that U.S. car sales hit a high in 2015. "Even though there's all this talk about ridesharing, people are still buying cars," he says. "I don't really worry about it, and don't let the team worry about it. I need the team focused on what we're doing."

What Arison and his team are doing has the potential to become a profitable, online used-car business. What he's not doing, again, is building an Uber.