Latest / The Operations Podcast with Fexingo: Process, People, and Profit in Modern Business / How Amazon Built a Last-Mile Operations Machine
Transcript
- Lucas: You order something on Amazon at 8 AM, and it's on your doorstep by early afternoon. That's not magic—that's a meticulously engineered last-mile operations machine. Luna: And it's something we've all come to expect, but the operational complexity behind it is staggering. Lucas: Exactly. So today we're going to unpack how Amazon built its last-mile delivery network. It wasn't always this way. For years, Amazon relied on UPS and FedEx to get packages to customers. But as Prime grew, those carriers couldn't keep up with the volume or the speed Amazon wanted. Luna: Right, and the cost was becoming a huge line item on their balance sheet. Lucas: If today's conversation gives you a new appreciation for the box that shows up at your door, the way these episodes stay ad-free is through listener support. You can buy me a coffee dot com slash fexingo. Now, back to the operations—Amazon started building its own delivery capacity around 2014, first with a program called Amazon Flex, where regular people use their own cars to deliver packages. Luna: That's the Uber model for packages. Lucas: Exactly. And then in 2018, they launched the Delivery Service Partner program, or DSP. That's where entrepreneurs start their own delivery companies, lease Amazon-branded vans, and hire drivers. Today, there are thousands of DSPs worldwide, and they deliver the majority of Amazon's packages. Luna: But how do they make that profitable for Amazon? I mean, running a fleet of vans, paying drivers, managing fuel—that's expensive. Lucas: The key metric is cost per package. Amazon has driven that down by optimizing route density. When you have a high volume of packages going to the same neighborhood, you can deliver more per hour. The magic number for a driver is about 40 packages per hour. That's the breakeven efficiency point where the economics work. Luna: So it's all about clustering orders geographically. Lucas: Right. Amazon's algorithms do that in real time. When you place an order, the system looks at all other orders in the area, groups them, and assigns them to the nearest delivery station. That station then sorts packages by route. Drivers get a pre-sorted route that minimizes drive time between stops. Luna: And that's why if you live in a dense urban area, you can get same-day delivery, but if you're in a rural area, it might take two days. Lucas: Exactly. Density is everything. Amazon's model works best in cities and suburbs. For rural areas, they still rely on the Postal Service or UPS for the last leg. But they've also been testing drones for those low-density routes. Luna: Drones—talk about a whole other operations challenge. But let's stick with the van-based system for a minute. How does Amazon manage the quality and reliability of all these independent DSPs? Lucas: They use a carrot and stick approach. DSPs are rated on metrics like on-time delivery rate, customer satisfaction, and driver safety. Top performers get bonuses and more delivery volume. Underperformers can get cut. Amazon also provides software tools for routing, tracking, and performance monitoring. And they handle the branding and the vans, so the DSPs just need to manage people and execute. Luna: It's a franchise model, essentially. Lucas: Very much like a franchise. But there's a twist: Amazon also runs its own delivery stations and hubs, so they control the sorting and the first mile. The DSPs handle the last mile. That gives Amazon flexibility—they can scale delivery capacity up or down without owning a huge fleet of vans and full-time employees. Luna: But isn't there a labor controversy? Drivers have complained about working conditions, and there have been unionization efforts. Lucas: Absolutely. That's a real tension. Amazon argues that DSP drivers are employees of the DSP, not Amazon, so they're not responsible for wages or working conditions. Critics say Amazon exerts so much control over the DSPs that they should be considered joint employers. It's an ongoing legal and regulatory issue. But from a pure operations perspective, the model is incredibly efficient. Luna: So what's the impact on Amazon's bottom line? Can we quantify how much the last-mile network saves them compared to using FedEx or UPS? Lucas: Amazon doesn't break out that number specifically, but analysts estimate that Amazon's cost per package for last mile is about 20 to 30 percent lower than what they'd pay to a third-party carrier. In 2025, Amazon delivered over 6 billion packages through its own network. If you save even a dollar per package, that's billions. Luna: And that savings can go back into funding things like faster Prime shipping or lower prices. Lucas: Exactly. It's a virtuous cycle. The faster and cheaper Amazon can deliver, the more people order, which increases density, which lowers cost per package further. That's the flywheel Jeff Bezos always talks about. Luna: But what about competitors? Can Walmart or Target replicate this? Lucas: They're trying. Walmart has its own delivery network called Spark, where gig drivers deliver orders. But they don't have the same package volume as Amazon, so their density is lower. And they don't have the same sortation infrastructure. Amazon has over 400 delivery stations across the US. Walmart is building out similar facilities, but they're years behind. Luna: And what about the environmental cost? More vans on the road means more emissions. Lucas: That's a fair point. Amazon has committed to net-zero carbon by 2040 and is rolling out electric delivery vans from Rivian. They've deployed over 100,000 electric vans already. But charging infrastructure and battery range are still constraints. It's a work in progress. Luna: So where does Amazon go next with last mile? Drones were mentioned, but what about autonomous delivery robots? Lucas: They're testing both. In some college towns, you'll see Amazon Scout robots rolling along sidewalks. And their Prime Air drone program has made limited deliveries in select cities. But the real next frontier is same-day delivery for even more items. Amazon is building smaller 'same-day delivery' facilities closer to customers, stocked with the most popular items. That way, you can order a laptop charger at 10 AM and get it by dinner. Luna: It's almost like they're turning every neighborhood into a mini fulfillment center. Lucas: Exactly. And that requires incredibly precise inventory placement. Amazon uses machine learning to predict what people in a given area will order and stocks those items in the nearest facility. It's a whole layer of operations on top of the delivery network. Luna: So the last mile isn't really the last mile anymore—it's the last hundred feet, and it's becoming a competitive battleground. Lucas: Right. And the companies that master it will win e-commerce. The ones that don't will struggle to compete on speed and cost. Luna: So as a consumer, when I get that package in hours, I should think about all the algorithms, vans, and people that made it happen. Lucas: Absolutely. It's one of the most impressive operations machines in the world today.