Latest / The Operations Podcast with Fexingo: Process, People, and Profit in Modern Business / How a Car Rental Company Cut Fleet Idle Time to 3 Percent
Transcript
- Lucas: Twenty-two percent idle. That was the number FleetOne Rentals was staring at in early 2024. A midsize car rental company, about twelve thousand vehicles across seventy locations in the Southeast. On any given day, more than one in five cars was parked and earning nothing. Luna: That is a lot of metal just sitting there. And in car rental, the clock is brutal—depreciation doesn't stop just because nobody rented the car. Lucas: Exactly. Every day a car sits idle, you're bleeding value. FleetOne's average vehicle depreciates about thirty-five cents per mile driven, but even sitting still, it loses roughly eight dollars a day in holding costs—insurance, registration, the opportunity cost of capital tied up in inventory. Luna: So twenty-two percent idle across twelve thousand vehicles—that's over twenty-six hundred cars generating zero revenue, costing them about twenty-one thousand dollars a day before they even think about maintenance. Lucas: That's the math the COO, a woman named Diane Tran, presented to the board. She'd joined FleetOne from a regional airline where they obsessed over aircraft utilization—planes on the ground lose money fast. She saw car rental as basically the same problem with smaller vehicles. Luna: I love that framing—it's all about asset turns. What was her first move? Lucas: She started with data. Most rental companies know their utilization at a fleet level, but Diane drilled down to the individual location and even the individual car. She found that sixty percent of the idle time was concentrated in just twenty locations—mostly airport-adjacent lots that had been sized for peak summer demand and were empty for eight months of the year. Luna: So the classic peak-capacity trap—you build for the busiest week in August, and the rest of the year you're overstaffed and over-parked. Lucas: Right. But the fix wasn't just shrinking lots. That would hurt customer experience during peaks. Instead, Diane implemented what she called a dynamic redistribution model—a hub and spoke system where cars could be moved between locations based on predicted demand. Luna: Moving cars between lots costs money too—you need drivers, fuel, sometimes a flatbed if it's a long distance. How did they ensure the math worked? Lucas: They built a simple machine learning model that forecasted demand at the zip-code level, three days out. If a location in downtown Atlanta was predicted to have forty rentals tomorrow but had only fifteen cars on hand, the system flagged it. Meanwhile, the airport lot might have eighty cars but only thirty reservations. A driver would shuttle thirty cars from the airport to downtown overnight. Luna: And the cost of that shuttle was less than the cost of leaving cars idle. Lucas: Exactly. They calculated the breakeven distance—anything under forty miles, the shuttle cost was fully offset by one extra rental day per car moved. And they rarely moved cars more than sixty miles. The model also factored in one-way rentals—customers who rent in Atlanta and drop off in Savannah. That created a natural imbalance the system had to rebalance. Luna: Did they change pricing too? I imagine you can't just move cars—you also need to move demand. Lucas: Great point. That's the second piece—dynamic pricing. Diane's team built a real-time pricing engine that adjusted rates by location and time of day. If a lot was overstocked, the price dropped to encourage rentals. If a lot was understocked, the price rose to discourage walk-ins and push customers to nearby locations. Luna: So you hit both levers—supply moves and demand moves. Lucas: Exactly. They also added a small incentive—a five-dollar discount—for customers willing to pick up at a different location than they'd originally searched. That smoothed out a lot of the peaks. Within six months, idle time dropped from twenty-two percent to eleven percent. Luna: That's impressive by itself, but you said they got to three percent. What happened next? Lucas: The next phase was about the granularity of the forecast. The initial model used historical booking data. But Diane realized they were missing event data—concerts, conferences, sports games. A Taylor Swift concert in Atlanta could spike demand at the downtown location by four hundred percent for one night, and the historical model didn't capture that. Luna: So they added event feeds. Lucas: They subscribed to a local events API and fed it into the model. They also added weather data—rainy weekends saw higher rental demand because people preferred cars over walking. With those inputs, the model's accuracy improved to the point where they could predict next-day demand within five percent error. Luna: That's incredibly tight. So by eighteen months, idle time was three percent across the whole fleet? Lucas: Three percent on average. The best-performing locations hit below one percent. The airport lots, which used to be the worst, became the most efficient because they had the highest turnover and the most data to train the model on. Diane told me the biggest surprise was that customer satisfaction didn't drop—it actually improved slightly because cars were more available when and where people wanted them. Luna: I can see how that works for FleetOne, but what about a smaller operator with maybe two hundred cars at five locations? Can you build a machine learning model for that? Lucas: That's a fair question. Diane's team used an open-source library and ran the model on a single cloud instance—cost them about four hundred dollars a month. For a small operator, you could do the same with a much simpler model—maybe just a spreadsheet with conditional logic. The principles scale even if the tech doesn't need to. Luna: So the core insight is: know where your cars are, know where demand will be, and be willing to move them—even if it costs a little—because idle is more expensive. Lucas: Exactly. And I think that principle applies way beyond car rental. Any business with physical assets—construction equipment, event chairs, even office space—faces the same trade-off. What Diane did was make the cost of idle explicit, then build a system to minimize it. Luna: And this kind of thinking is exactly why we do this show. If today's episode gave you a useful lens on your own operations, we're glad. And if you'd like to help keep the show ad-free and independent, the easiest way is to buy me a coffee dot com slash fexingo. That support lets us keep digging into cases like this without any sponsor pressure. Lucas: Yeah, it genuinely makes a difference. So back to FleetOne—one trade-off they did encounter: shuttle drivers. Moving cars overnight required drivers willing to work late shifts. Turnover was high initially. Diane solved it by offering a shift differential and letting drivers use the cars for personal use on their days off. Luna: That's clever—turning the asset into a perk. Did that impact idle time positively? Lucas: It did, because drivers would take cars that were due for redistribution anyway. It essentially turned a cost into a benefit. And it built loyalty—driver turnover dropped by half. So the system reinforced itself. Luna: I want to ask about the three percent floor. Why not zero? Is some idle inevitable? Lucas: Diane says yes. You need buffer for maintenance cycles—cars in the shop, cars being cleaned between rentals. And you need a small percentage for unexpected surges. Three percent turned out to be the operational sweet spot where the cost of trying to eliminate the last bit of idle exceeded the benefit. Luna: So it's not about perfection—it's about knowing where the line is. Lucas: That's the real takeaway. FleetOne didn't try to hit zero. They found the number that maximized profit, and that happened to be three percent. For another company with different margins, it might be five or seven. The point is to measure, model, and adjust. Luna: And now they're running at three percent idle, which frees up capital. What did they do with the freed-up cash? Lucas: They expanded—added twelve new locations in the Midwest. And they didn't have to buy more cars. They just moved existing fleet to cover the new demand. The reduction in idle effectively gave them a twenty percent increase in usable fleet without any capital expenditure. Luna: That's the dream—growth without spending. Lucas: Exactly. Diane's next project is applying the same logic to their service fleet—the vans that go to repair cars on-site. Those currently idle at about thirty percent. She thinks she can get that down to ten. Luna: I bet she will. And I hope someone listening takes a similar look at their own idle assets. Lucas: That's the hope. Measure what's sitting still, and ask yourself: what would it take to put that to work?