This article is from Full Stack Economics, a newsletter about the economy, technology, and public policy.
It was the Saturday night before Halloween, and I was ferrying costumed, intoxicated young adults around our nation’s capital. Shortly before midnight, I picked up a normally dressed woman in downtown D.C., who was heading home after a large fundraising dinner at the city’s convention center.
Demand was off the charts, and my passenger told me that Lyft had charged her $59 for our six-mile trip to Arlington, Virginia. I was surprised because that was far more money than I’d earned on any of the 24 trips I’d completed so far that week.
After I dropped her off, the Lyft app revealed my portion of the base fare: $16.52. My passenger added a generous $8.96 tip, for total earnings of $25.48. That made it my most lucrative trip that week. But Lyft raked in much more: $42.48.
A lot had changed since the last time I drove for Lyft as a journalistic experiment in 2014 (and wrote about it for Vox). Back then, drivers took home a standard 80 percent of a passenger’s fare, sometimes with tips and bonuses on top. Earlier this fall, I decided it was time to embed in the driver’s seat of a Lyft once again. It turned out that in recent years, the relationship between driver compensation and passenger fares has become … complicated.
After that Saturday night trip, I started asking my passengers how much they’d paid and comparing it to what I’d earned. The results were all over the map. I had two trips where I got more than 80 percent of the total fare, including tips. I had a couple of others where I earned less than 30 percent.
I wrapped up my Lyft driving experiment after a week, having completed 100 rides in 46 hours and earning $1,111. Lyft eventually sent me a report showing that passengers had paid a total of $2,139.73 for these rides, so on average I got just 52 percent of what my passengers paid.
Ever since then, I’ve been trying to figure out how Lyft could take such a big cut of passenger revenue and still fail to turn a profit—Lyft says that it lost almost a billion dollars in the first nine months of 2022. The more I’ve looked into it, the more I became convinced that it’s going to take major cost-cutting for Lyft to avoid eventual bankruptcy. I was also left asking one question: If after all these years, Lyft and its competitor Uber’s disruption of the old taxi model still doesn’t make business sense, what was the point?
An Overgrown Taxi Dispatch Service
There’s a thought experiment I find illuminating: Imagine that Lyft announced tomorrow that it was going to shut down its app and lay off most of its staff. Then it was going to open a bunch of call centers and hire people to book taxi rides over the phone instead.
When a passenger wanted a ride, they would call a toll-free number (1-800-GET-LYFT, perhaps) and provide pickup and dropoff locations. The operator would then relay this information to drivers, perhaps in a group chat or over an old-style radio system, and one of them would claim the ride. The operator would provide the passenger with an estimated pickup time and take credit card information over the phone.
Here’s the question about this hypothetical version of Lyft: Would it be more or less expensive to operate than the company that actually exists?
Intuitively, you’d expect the answer to be “more expensive,” right? After all, booking a ride on the Lyft app is an automated process. A single server in Lyft’s data center should be able to book hundreds if not thousands of rides in an hour, while costing less to operate than the wages of a call-center worker.
Yet there’s good reason to think that my retro version of Lyft would actually be much cheaper to operate than the high-tech version. I’ve basically described an old-fashioned taxi dispatch company—the kind that dominated the industry before Uber and Lyft came along. Back in 2016, a transportation economist named Hubert Horan crunched the numbers and found that in the pre-Lyft world, about 15 percent of passenger fares (including tips) went to dispatching and other “back office” functions. The other 85 percent went to driver compensation, fuel, and the costs of vehicle ownership and maintenance.
It seems that the smartphone revolution didn’t make taxi dispatching cheaper and more efficient. It made it way more expensive.
So if Lyft were as operationally lean as a traditional taxi company, the driver would get 85 percent of passenger fares (Lyft drivers pay to maintain and fuel their own vehicles, after all), while Lyft would be able to cover its costs with the other 15 percent. But if my rides are any indication, Lyft is taking way more than 15 percent and still losing a ton of money. It seems that the smartphone revolution didn’t make dispatching cheaper and more efficient, as you might expect. It made it way more expensive.
I’m leaning on my own experience here because Lyft is frustratingly opaque about its financials. Lyft publishes a net revenue figure with driver pay already subtracted out. As far as I can tell, Lyft doesn’t publish its gross bookings, making it impossible to calculate Lyft’s overall “take rate.” The company declined to speak to me for this story or provide data on how it splits fares with drivers.
Uber is more transparent. Last quarter Uber reported ride hailing revenues of $3.6 billion on gross bookings of $13.4 billion, for a “take rate” of 28 percent. That’s still a larger cut than traditional taxi dispatchers take, according to Horan’s math, but it’s much less than the 48 percent “take rate” I experienced as a Lyft driver.
“This Is a Crappy Business”
Last month I argued that rising interest rates were forcing Silicon Valley to undertake a significant culture change. During the 2010s, companies like Uber and Lyft focused on rapid growth without worrying too much about costs. With interest rates at rock bottom, investors were willing to throw billions of dollars at these companies hoping that they’d be the next Amazon or Facebook—companies whose early losses eventually turned into big profits.
Now, with growth slowing and interest rates rising, Lyft is running out of time. The company needs to figure out how to reach profitability at its current scale.
Horan has been arguing for many years that this is impossible because Lyft and Uber have not significantly improved on the business model of traditional taxi dispatchers.
“This is a crappy business,” Horan told me in a recent phone interview. “The only way Uber has ever improved its profit margins is by screwing drivers.”
Like any labor-intensive industry, taxi companies face a basic tug-of-war between workers and customers. Every dollar of driver income comes out of the pockets of passengers. With no real way to raise driver productivity (as might happen in a factory), the only way to raise driver income is by charging passengers more money. But most potential riders aren’t wealthy, so if you charge more, you’ll get fewer customers.
Five years ago, Uber and Lyft aimed to reshape the taxi industry by enabling carpooling at a large scale. Services like Uber Pool and Lyft Line really could have improved the economics of the taxi business, since it would have allowed a driver to earn income from multiple passengers at the same time. But my week as a Lyft driver made me skeptical these services will amount to much: only five out of my 100 rides were shared. That could partly be an aftershock of the pandemic crushing demand for shared rides, but it doesn’t seem like a promising sign.
I think it’s overstating things to say that Uber and Lyft haven’t improved on traditional taxis. At a minimum, Uber and Lyft have dramatically improved the reliability of pickups. I’m old enough to remember calling a taxi in the pre-Lyft days and not knowing if it would show up. I think many passengers are willing to pay at least a small premium for that.
Lyft’s software also has some features that boost driver productivity:
• Because drivers use their personal vehicles, they have the option to head out on the road during periods of peak demand, like Friday and Saturday nights.
• When I was driving, Lyft frequently had a new passenger ready for pickup before I’d even dropped the previous passenger off, minimizing waiting between passengers.
• At the end of a shift, Lyft has a feature that lets drivers only accept rides that will get them closer to home. This helps to avoid “deadheading,” where the driver has to waste time and fuel driving home with an empty car.
In theory, conventional taxi dispatchers could do those last two things, but I bet the combination of large scale and precise GPS coordinates enables Lyft to do it more effectively.
But with that said, I think Horan is right that these are fairly incremental improvements to the basic taxi business model. Which leaves Uber and Lyft with the same challenging economic model as a traditional taxi company.
Journey to Maryland
The longest and most interesting trip of my week as a Lyft driver came on Sunday, Oct. 30 around 10 p.m. I picked up a young rapper named Sir E.U in D.C.’s Adams Morgan neighborhood who asked if we could listen to his music on my car’s speakers. With the music playing, we proceeded to Children’s Hospital, where I picked up a second passenger I’ll call Michael. We then drove south into Maryland, dropping off Sir E.U, and then continuing to drop off Michael in Waldorf, Maryland, almost an hour’s drive south of the hospital.
During our long drive south, Michael told me that he’d just left his one-year-old daughter with her mother so she could get surgery for an unusual growth on her neck. He would have liked to stay the night, but he needed to get back to his low-paying retail job on Monday morning.
Michael was 21 years old and told me that he’d been working as a plumber’s apprentice until his car broke down. For a while, he tried to continue his training by taking Lyft rides, but after racking up an $800 Lyft bill in one month he realized he couldn’t keep doing that. So he was stuck in a dead-end job until he could save up enough money to get his car fixed.
Michael paid $61 for his trip to Waldorf, and Sir E.U paid another $19.99, for a total of $80.99. I earned less than half of that—$38.72—for the ride.
If you’re a college-educated professional, you probably take Lyft for a night on the town or to get to the airport. You might assume most of Lyft’s passengers are like you. And I certainly did have a fair number of passengers like that.
But I also had a lot of passengers like Michael: lower-income people who didn’t have a car and relied on Lyft to get places transit couldn’t take them. Michael clearly can’t afford to take $61 taxi rides on a regular basis, but the situation left him with no good options.
It’s passengers like Michael—even more than Lyft drivers—who would benefit if Uber and Lyft figured out how to make their dispatching service more efficient. Lyft effectively charged Michael around $30 to arrange our ride. I bet if they put their minds to it, they could figure out how to do the job for much less than that. That’s not because it actually costs $30 to connect a rider with a driver—it’s because Lyft needs billions of dollars to cover its massive overhead.
The Case for Frugality
There’s a wide range of attitudes toward frugality within the tech industry. On the one hand, you have a company like Google that provides its employees with free meals and every imaginable perk. Google has spent billions of dollars on “moonshot” experiments like self-driving cars with little regard for their costs. Google is able to do this because its search engine is a quasi-monopoly that throws off billions of dollars in profit every year.
Google’s workplace culture has had a big influence across Silicon Valley—including at Lyft, which like Google has long offered employees free lunches.
But not every company works like Google. At the other extreme you have a company like Amazon that builds desks out of doors to make a point about the importance of thrift. Amazon had to develop a frugal culture in its early years because it was operating in the low-margin retail industry, competing with famously tight-fisted companies like Walmart.
I think a lot of tech companies, including Lyft, could stand to be less like Google and more like Amazon. And that’s especially true for a company like Lyft that is (like Amazon) operating in a low-margin, labor-intensive industry.
Back in April, venture capitalist Marc Andreessen wrote that “the good big companies are overstaffed by 2x. The bad big companies are overstaffed by 4x or more.”
Then last month, Elon Musk put that theory to the test. He laid off around two-thirds of Twitter’s 7,500 employees, leading to numerous predictions the site would “break.”
Twitter has obviously had plenty of problems in the last six weeks, but they haven’t included large-scale outages. Twitter’s engineers seem to have been able to keep the site running with a much smaller staff.
And that makes me wonder whether something similar might work at Lyft. Last month, Lyft laid off 13 percent of its employees, leaving it with a workforce of around 5,000 people. Could Lyft make much deeper cuts while keeping the core business running? It’s hard to know from the outside. But I hope the company’s leadership gives it a try. Because dramatically lower overhead would leave more room for Lyft to offer lower passenger fares and higher driver pay—in other words, to prove that all that disruption had a purpose.