In addition to being a political scientist focusing on post-Soviet transition, George Arison has been an innovator in transportation. He founded Taxi Magic, an on-demand transport company far ahead of its time, and is now focused on Shift, which introduces a new model for buying and selling used cars.
Elliott Adams: Taxi Magic is especially interesting because you guys were ahead of the curve, right? You were ahead of this idea of on-demand transportation.
George Arison: Yeah.
EA: What's the origin story of that company?
GA: I was traveling a ton as a BCG consultant, and we'd take taxis all the time because I didn’t have a driver's license, and so when you would arrive in some random town or city and try to take a cab to your client, to get up here, that was really too much. So, I wondered if you could use an app on a phone like a Blackberry to book a taxi.
I had a mentor and a friend named Tom DiPasquale, who had started a company called Cookbook, which was then bought by what was then Concur. And as a result of that, Tom's technology was used by Concur as the primary way to do all the travel bookings and expense management. What Tom was saying was that the vast majority of expense that is done in cash is ground transportation, and it's a very idiotic situation because it's where a lot of the fraud happens. So, he was really into this idea of whether you can use mobile in some way to create a more electronic receipt and electronic payment trail for ground travel. Those two ideas kind of coming together is what started Taxi Magic.
I think one of the things was we fundamentally thought that we knew what the right answer was when we got started, and that was not the right thing. In terms of how to approach it, we should have learned our way into the product a lot more than we did, and I think that's something really critical. We wasted a ton of money and a ton of time on stuff we thought was right, but then it turned out the product was being used very differently or features were not being used at all. Now, granted, agile development and the user feedback method was not as prominent back then as it is today.
So, that's number one, and number two is that we didn't focus enough on building a strong enough technology stack to make all this work. As a result of that, our ability to innovate when we were being out-innovated by someone like Uber was a lot slower. Thirdly, I think we didn't go full stack because we wanted a simple software product, versus Uber and Lyft, who are not a simple software product. I think that was a mistake because we were very dependent on taxi fleets to provide the service, and they were doing a pretty terrible job at that. That’s kind of the three big things I tend to usually focus on in terms of lessons.
EA: I assume some of those lessons carried over to what you did with Shift.
GA: For sure, and in some ways, we may have over-corrected at Shift a little too much in some cases and now are correcting ourselves because that's just the reality. The answer is never as cut and dried one way or the other. But yeah, Shift was built in a very different way. Before the company was even started, I spent a ton of time testing the initial product idea and discovered that what we initially were thinking was going to be the product was actually not going to work because the consumers didn't think about that product first.
So, we started out by really being interested in lending and how to enable lending in non-dealer transactions when cars are being sold or bought, the reason being that it was a pain point personally for me, and I had lived through that experience. Secondly, it seemed like a huge market opportunity because no one else was doing that. But it turns out that there's a reason why it's like that. People don't think of the loan first, and those that do are not the customers you want to offer a loan to anyway since they're the least likely to pay their loan off. We tested our way into knowing those things and that it was not the right product before we had any tech team or a business or a company incorporated in any of that. So, we really redid the product through user testing really early, and that generally has been the approach that Shift has taken to product development.
EA: How did the customer experience evolve? My understanding is that you send out a rep, is that correct?
GA: Initially we didn't, but in the beginning, the thought was to not even have any touching of the car. That all the Shift product would be around the lending and the warranty and offering a guarantee on a car. Say you had a car you were selling, and Jamie wanted to buy that car. Shift would inspect your car, certify it, attach a warranty to it as a guarantee. You would still be responsible for selling the car, and you would still be responsible for finding Jamie as a buyer, but then when Jamie felt ready to buy this car, through Shift, you'd be able to offer financing to Jamie to buy that car. That was the initial product idea, and then the more we spent time with users, it became clear that the users actually wanted the car off their hands, and they wanted a more augmented set of services than the ones that we offered them. They kind of wanted a much more Uber regular or UberBLACK experience, versus the UberX experience, so that analog is number one.
And then number two is, we then had this idea of a test drive delivered to you, which is kind of how we got started. That evolved over time, as well, so we initially were sending out a very expert person to do the test drive. This was a highly paid, very knowledgeable person who knew a ton about cars, who'd come out to your house or your office or whatever. That was a very good experience for customers, but we found we would run into a whole bunch of different issues.
Number one, it was very difficult to optimize their time because they would spend so much time on the road, and you could maybe fit in two test drives a day, but you could never break even on two test drives a day. It was too expensive. Secondly, it was very hard to train them because they were on the road so much, and so the variability between the high-quality performance and the medium-quality performance was so high that it just didn't work, and a medium performer would drive down the average for everybody.
So then our president, the man who started Taxi Magic with me, Toby Russell, said, “Okay, what can we do to transform this to a place where we keep this unique and kind of secret sauce of test drive delivered to the customer, but do it in a more optimized manner where the costs are not as crazy and the quality of the customer experience is not suffering because of the difference in the quality of people providing the service?”
So, we tested our way into a model where we have a person on the phone who is the expert, like a sales rep, basically, and then a driver, who is just the regular driver, delivers the car to the customer for a test drive. So, now the customer gets the test drive experience but also is getting an expert. He just never meets the expert, so the expert's not spending time on the on the road anymore, and instead is able to help many more customers by phone. The end result of that has been that our conversions on test drives to sales have gone up by over 10 points. Before, we'd be converting 32–34 percent test drives to sales conversions. Now we're in a kind of a 42–45 percent range.
GA: Number two, our customer feedback actually has gone up as well, because now everybody is getting the best person, and it's easier to train everyone because they're all in one office at the same time. So, our costs have come down because we can optimize schedules, and our utilization is way better this way than we were able to do in the old model. Now, we tested our way into doing that where we had this idea; we did it for about five percent of customers, then we grew it to fifteen, then we grew it to twenty, then we tweaked it. We concurrently built technology to support it, etcetera, etcetera, etcetera, and over the course of Q2 and like, Q3 and Q4 last year, we made this huge change, and in January, converted the business to this model. So, that is definitely a very different way of doing things than Taxi Magic used to, and I think it's a more effective way to be successful with building a business.
EA: Could you dive into that a little more? For Shift, how did you and your investors come to define product/market fit?
GA: That's a tough question. There’s a Supreme Court case in which Justice Scalia writes this great line, “You'll know it when you see it.” It applies to this as well. You have product/market fit when you see it, and you don't when you don't see it. There's not an ironclad rule about that.
This business is different than a lot of businesses because the value of the transaction in our space is so massive, right? We make $3,500 a car sold in terms of revenue when we do a transaction. Think about how many Uber rides that is. Uber makes maybe about a dollar fifty per ride, so you're talking about an insane number of Uber rides—about 2,300 Uber rides to make up one car revenue. And so, you don't have the luxury of lots of volume to judge things because it's just not possible to get to that kind of volume without additional capital. But I think you wonder if it’s working. For us, the single biggest thing was people giving us cars. It was very clear, obviously, you could sell a car. Forty-five thousand cars are sold a year, but the idea that a person would give you a car for free to sell and you don't pay them money up front is a pretty big deal, in my opinion. Once we were proving that we were actually doing that, that's when I think people felt like we had product/market fit.