Latest / The Operations Podcast with Fexingo: Process, People, and Profit in Modern Business / How UPS Used ORION to Save 100 Million Miles
Transcript
- Lucas: UPS drivers make roughly 160 to 180 stops per day. Each stop is someone's front porch or loading dock. And for decades, the routing decisions — which house to hit first, which way to turn — were left entirely to the person behind the wheel. Luna: Right, the classic 'driver knows the route' model. And for a long time that worked pretty well. Lucas: It did. Until the numbers got too big. By the early 2000s, UPS was running 55,000 delivery vehicles across the U.S., burning through a billion miles a year. A single percentage point of inefficiency meant millions of dollars in fuel and labor. Luna: So they built a computer to out-optimize every driver. That's ORION, right? On-Road Integrated Optimization and Navigation. Lucas: Exactly. ORION is a routing algorithm that takes in every package on a truck — its size, its destination, the delivery time window — and computes the optimal sequence of stops. But the real breakthrough wasn't the math. It was getting 55,000 veterans to follow it. Luna: Because a driver who's worked the same route for fifteen years has a mental model of the day that the algorithm doesn't share. Lucas: Exactly. The algorithm wanted to minimize left-hand turns. UPS had data showing that a left turn across traffic costs roughly thirty seconds of idle time and half a cent of fuel. Multiply that by millions of turns and you're talking real money. But drivers hated being told not to turn left. It felt unnatural. Luna: And ORION's first rollout in 2012 was pretty rocky, if I remember. Drivers complained the routes took longer than their own. Lucas: They did. UPS had to pull the system back, recalibrate, and spend two years embedding the algorithm's logic into training. They also added a feedback loop — drivers could submit exceptions. If you knew that a certain road flooded after rain, you could flag it, and ORION would learn. Luna: So they treated the driver as a sensor, not a problem. That's smart operations thinking. Lucas: It's the only way massive optimization projects work. You can't mandate a top-down system on a workforce that has decades of tacit knowledge. You have to absorb that knowledge into the algorithm. UPS called it 'guided optimization' — the computer suggests, the driver decides. Luna: And the numbers they eventually published are staggering. By 2020, ORION had saved 100 million driving miles and 10 million gallons of fuel. Annual savings around $300 to $400 million. Lucas: The project cost about a billion dollars to build and deploy across the fleet. So it had a payback period of roughly three years. That's a solid return for an internal software play. Luna: But the interesting thing is that ORION isn't a static program. It keeps getting updated with real-time traffic data, weather, even package dimensions now. Lucas: That's the part I want to drill into. The 2022 version, ORION 3.0, started incorporating telematics from the trucks themselves — engine idle time, braking patterns, fuel consumption. So now the algorithm isn't just routing you efficiently; it's coaching your driving behavior in real time. Luna: That's a lot of data per driver. Does that cross a line? Lucas: It's a fair question. UPS frames it as coaching, not surveillance. The driver sees their own score compared to the fleet average. There's no punishment for a high idle time — but the system flags it, and the driver can adjust. And the union, the Teamsters, has negotiated that this data can't be used for discipline. It's purely for efficiency. Luna: Still, I'd be annoyed if my truck was telling me I brake too hard. Lucas: Some drivers probably are. But the company has been careful. They introduced the telematics feature as a voluntary pilot first. Only after three years and positive feedback did they roll it fleet-wide. Luna: Smart rollout strategy. So ORION's story isn't really about the algorithm. It's about change management at massive scale. Lucas: That's the takeaway I keep coming back to. A lot of companies try to bolt optimization software onto an existing workforce and expect it to stick. UPS spent almost a decade iterating on the human side before the algorithm was fully trusted. Luna: And they had the advantage of a unionized workforce that forced them to negotiate rather than dictate. That actually made the system better. Lucas: I think so. The Teamsters demanded that any routing change had to be proven to improve safety as well as efficiency. That forced UPS to build ORION with a safety overlay — no route that required a U-turn on a highway, for example. Luna: It reminds me of Toyota's andon cord system, where any worker can stop the line. The principle is the same: the person doing the work has authority to override the system. Lucas: Exactly. And that brings us to a broader point about operations. The best processes are the ones that combine computational optimization with human judgment. ORION gave drivers a 95 percent optimal route and let them use the last 5 percent for local knowledge. Luna: Speaking of the show — we don't run ads, and we want to keep it that way. If today's episode was useful to you and you want to support that, buy me a coffee dot com slash fexingo helps keep these conversations going. Lucas: Yeah, it literally funds the research and production time. Appreciate anyone who chips in. Now back to ORION — one metric I find fascinating is the left-turn reduction. UPS claims that by 2025, they'd cut left turns by 90 percent on optimized routes. Luna: Ninety percent? That's almost total elimination. How do they even deliver packages without turning left? Lucas: They use a three-right-turn loop. If you need to go left, you go around the block — right, right, right. It adds distance but saves time because you're not waiting for a gap in oncoming traffic. Luna: And the fuel savings from reduced idling more than offset the extra miles. Lucas: Exactly. UPS calculated that each left turn avoided saves about 0.4 percent of a mile's fuel. Doesn't sound like much, but over a hundred million miles, it's huge. Luna: So the math works. But I wonder about scalability. UPS is a single company with a single culture. Can other fleets replicate this? Lucas: Several have tried. FedEx has a similar system called FOCUS. Amazon uses routing algorithms for its delivery service partners. But the ORION project is uniquely well-documented because UPS published case studies and academic papers. It's become the textbook case for route optimization. Luna: What about smaller fleets? Like a regional bakery with twenty trucks. Lucas: They can buy route optimization software off the shelf now. Companies like Routific and OptimoRoute offer it as a service for a few hundred dollars a month. The barrier is no longer technology — it's getting the drivers to trust it. Luna: So the UPS lesson applies at any scale. The rollout process matters more than the algorithm. Lucas: Exactly. You need a feedback loop. You need to let drivers override when the system is wrong. And you need to show them the data — 'look, your route used 3 percent less fuel today.' When drivers see it working, adoption follows. Luna: And the environmental angle is real. UPS's 2025 sustainability report credited ORION with avoiding 100,000 metric tons of CO2 emissions annually. Lucas: That's the equivalent of taking 21,000 cars off the road. And it came from a software project with a three-year payback. That's the kind of story that should make every logistics manager rethink their routing. Luna: One last question — is ORION still improving? Or has it plateaued? Lucas: It's still improving. UPS recently added predictive analytics for package volume. If the algorithm knows that a certain neighborhood gets fifteen packages every Tuesday in December, it can pre-position the truck closer to that area. That's a level of granularity that wasn't possible five years ago. Luna: So the frontier is moving from static route optimization to dynamic, real-time fleet orchestration. Lucas: That's exactly where it's headed. And if the trend continues, the next generation of ORION won't just tell drivers where to go — it will coordinate the entire delivery network in real time, rerouting trucks based on traffic, weather, and even e-commerce order spikes. The 100-million-mile savings might just be the beginning. Luna: That's a powerful vision. Thanks for unpacking the story. Lucas: Thanks for listening. We'll be back next week with another deep dive into how operations shape the businesses we rely on.