3D Robotics

The economics of drone delivery


Does drone delivery make economic sense? The answer, in this good analysis at Flexport by Dan Wang, is that it certain limited conditions with the right combination of market density or poor existing infrastructure, it might. Excerpt:

Drone deliveries look like the future: unmanned quadcopters rapidly delivering packages to our doors, eliminating both wait times and the cost of human labor.

But from an economic perspective, it’s easy to see how drone delivery could be an elegant technological solution in search of a problem.

That’s because the economics of last mile delivery are driven by two factors, route density and drop size. Route density is the number of drop offs you can make on a delivery route, often called a “milk-run” in industry parlance. Drop-size is the number of parcels per stop on the milk run.

If you make lots of deliveries over a short period of time or distance, the cost per delivery will be low. Likewise, if you drop off lots of parcels at the same location, the cost per parcel will be low.

Drones perform poorly on both of these economic aspects of last-mile delivery. The current prototypes that companies have unveiled usually carry just one package, and after the drone makes its delivery, it has to fly all the way back to its homebase to recharge its batteries and pick up the next package.

Compare that to the current status quo: delivery trucks. A delivery truck from UPS makes an average of 120 stops a day to deliver hundreds or thousands of packages. Don’t they seem to be better than drones?

How short-distance drone deliveries work

In late November, Amazon released a slick video demo of Prime Air, a drone delivery system designed to “get packages to customers in 30 minutes or less.” It comes on the heels of a similar production from Google’s Project Wing, which showed a drone delivering dog food in Queensland, Australia.

Both these companies and others say they expect to be flying soon.

We approached these companies’ claims with skepticism. After all, in Amazon’s demo, a drone that the company says can fly 15 miles delivers a pair of soccer shoes. But what if you don’t live within 7.5 miles of an Amazon warehouse? And will Amazon keep drones on standby for when you order 50 pounds of diapers? Drone delivery is still speculative, and Amazon and co. aren’t revealing all their plans. But a few simple statistics show that distance and weight may not hold back drone delivery.

In official documents, Amazon has written that 86% of its packages weigh under 5 pounds. As for distance, Walmart has noted that 70% of Americans live within 5 miles of a Walmart. This is great for the prospect of drone delivery. Amazon’s fulfillment centers are not as ubiquitous, but the company has shown willingness aplenty to move its products and warehouses closer to the customer.

If products start close enough to consumers that drones can deliver them in 30 minutes, we may very well see drone delivery even if it costs more than the humble UPS truck. “Faster delivery has always been a goal,” says Logan Campbell, the CEO of Aerotas, a drone consultancy. “Seven day delivery was common. Now we’ve become used to Amazon Prime 2-day delivery.”

There haven’t been many analyses of drone air freight costs. The ones that do exist suggestion that drones have the potential of being both faster and cheaper than delivery costs.

In a report by ARK Invest, Tasha Keeney suggests that Prime Air could cost Amazon only 88 cents per delivery. If Amazon charged customers $1 per delivery, Keeney estimates, the company could earn a 50% return on its investment in drone infrastructure while offering same-day delivery that is significantly cheaper than current alternatives.

The analysis is still mostly speculative. Keeney imagines that 6,000 operators who earn $50,000 per year will operate 30,000 to 40,000 drones. Each drone will make 30 deliveries per day. Her analysis ignores depreciation, and questions like: “How will drones avoid airplanes and deliver packages in Manhattan?” And there’s another core issue: $12.92 is the price UPS charges to consumers, but its actual marginal cost of delivering one more package along a route they are delivering to already is probably closer to $2. When push comes to shove, will drones be able to compete? The rest of her analysis incorporates the costs of electricity, backup battery packs, bandwidth, upgrades to facilities, and so on.

Keeney’s assumptions also stick to the middle ground: She presumes that Amazon will gain permission to fly drones out of sight, with each operator responsible for 10-12 drones, but not that Amazon will soon automate the entire process. That said, other analyses, which assume that more pilots will be needed, put the cost closer to $10-$17 per delivery.

Delivering blood samples in Lesotho

An experiment from the other side of the world offers some validation of ARK’s rosy prognosis.

The drone startup Matternet is not just planning a drone delivery network. It has already run one—in Lesotho, a landlocked country surrounded by South Africa.

Matternet has delivered crucial supplies (and chocolate) via drone in Haiti after the earthquake, and when the founders wanted to prototype a drone network, they turned to Maseru, the capital of Lesotho. In Lesotho, almost 1 in 4 adults has HIV, and even in the capital, paved roads are scarce, which makes it difficult to transport goods. So Matternet’s drones delivered blood samples from clinics to hospitals where they could be analyzed for HIV/AIDS.

Blood samples were perfect cargo: small, light, valuable, and time-sensitive. Since Maseru has little air traffic and the routes from the clinics to the hospitals did not change, most of the process could be automated. The drones flew without a human pilot and had clear landing areas where they recharged automatically. Matternet CEO Andreas Raptopoulos says it took their drones 15 minutes to fly 4.4 pounds of cargo 6.2 miles, and that the Maseru network successfully covered an area 1.5 times the size of Manhattan.

It was a good price point. Even though the pilot took place several years ago, Raptopoulos saysthat each delivery cost only 24 cents.

Test cases for drone deliveries

The first legal delivery in the United States via drone took place on July 17, 2015.

That day, a drone operated remotely made three trips to transport medicine from the Lonesome Pine Airport in Wise, Virginia, to a nearby fairgrounds. The demonstration was the result of a partnership between the drone startup Flirtey and 2 organizations that provide healthcare in rural, underserved areas.

The flight demonstrates two aspects of the future of drones and air freight: that technology is not the limiting factor, and that drones’ most obvious appeal is not for personal deliveries.

“The technology is here,” says Logan Campbell of Aerotas. “You see fully autonomous flights already” for uses like surveying construction sites, farmland, and mining operations.

“It’s gotten incredibly easy to fly,” adds Roger Sollenberger, spokesman of 3D Robotics, whose free, open source software powered Flirtey’s delivery drone. 3D Robotics’ drones, he says, make it so easy to program flight paths that the average parent can use it to record family moments.

Instead, the real challenge is the regulatory environment. The FAA has banned all commercial uses of drones in the U.S., and while the agency increasingly grants exemptions, the Flirtey flight is the only freight exemption that allowed a real delivery rather than testing in unpopulated areas.

The experts we spoke to expressed cautious optimism that the FAA, which is working on guidelines for drone deliveries, will let them fly. The FAA currently requires companies with exemptions, like Amazon, to have an operator with a pilot’s license keep each drone within line of sight—a mandate that makes deliveries completely uneconomical.

Can drones substitute roads?

In the meantime, the first drone deliveries will probably look more like Matternet’s drone network in Lesotho and Flirtey’s medicine drop in Wise, Virginia. It’s much easier to operate and gain legal permission to fly in open skies, and the economics are particularly compelling.

As Raptopoulos of Matternet points out, Google and Amazon’s plans ignore drones’ best feature: they can go where there are no roads.

“One billion people in the world today do not have access to all-season roads,” Raptopoulos tolda TED audience in 2013. “We cannot get medicine to them reliably, they cannot get critical supplies, and they cannot get their goods to market in order to create a sustainable income.”

For the Matternet team, the most interesting question was not the cost per delivery. They wanted to compare the cost of the drone network to the cost of building the roads Lesotho so badly lacks.

The American Road and Transportation Builders Association estimates the cost of a “new 2-lane undivided road” at $2 million to $5 million per mile. And the drone network in Maseru? “Less than a million dollars,” Raptopoulos said in 2013.

Comparing a drone network to roads that can transport buses, delivery trucks, and bulldozers is not an apples to apples comparison. But given that building and maintaining roads is a long and expensive process, drones could offer a quick and cheap way to (imperfectly) connect the billion people identified by Raptopoulos as cut off from most of the world.

DHL-deutsche-posthttps://www.flexport.com/blog/wp-content/uploads/2015/12/DHL-deutsche-post.jpg 667w" sizes="(max-width: 667px) 100vw, 667px" />

A drone delivery test by German Deutsche Post AG in December 2013. Photo by Frankhöffner

“Many in the industry think delivery will take a different feel from Amazon,” says Campbell of Aerotas. “More specialized cases like delivering vaccines that need to be refrigerated to regional hospitals… Niche deliveries where speed is critical.”

In these cases, for cargo that is small, light, valuable, and time-sensitive, cost is much less of a factor. A drone delivery may save a life by getting delicate medicine to a rural patient, or keep an oil rig running by delivering a key piece of machinery.

Even in less extreme cases, drones are appealing because route density and drop size are less relevant. Instead, drones just need to beat the cost of private couriers, which cost at least $10 to $50, even within dense urban areas like San Francisco.

Hospitals that need delicate medicines. Oil rigs that need replacement parts. Residents in remote Alaskan towns that need key supplies. Companies that need timely data about their supply chains. Expect these (noncommercial) efforts to be the early adopters of drones and drone delivery.

When is the future?

Drones are in a situation similar to the one faced by self-driving cars. Companies have demoed the technology, so the real obstacle is the legal and regulatory environment. In both cases, this means integrating the technologies into daily life could take a long time—or it could happen very quickly.

Current drone economics are good for deliveries that take less than 1 hour. Are people willing to pay a major premium for the service? Perhaps. It may not be too late to order Christmas presents on December 23, 2020.

Despite drones’ current inability to match the efficiency of a delivery truck’s milk run, the economics of delivering air freight by drone seem compelling. That’s why Amazon and Google are investing in the R&D. That’s why Matternet is testing drone deliveries with Swiss Post and Swiss World Cargo. And that’s why the drone community expects deliveries to happen—even if not as quickly as executives like Bezos promise.

In the meantime, drone deliveries will probably get their start in remote areas like Lesotho, or in use cases like flying vital machine parts to oil rigs and mines, or by collecting data on shipping to make it more efficient.

E-mail me when people leave their comments –

You need to be a member of diydrones to add comments!

Join diydrones


  • I think drone delivery won't be fully viable until battery technology undergoes a revolution.

    Also, I think drone delivery is going to play out a lot different to what people think. At the moment everyone is focused on delivery from a warehouse to the customer via drone.
    That only works in a limited set of circumstances.
    Getting rid of the last mile for deliveries is a lot easier when you incorporate it into the long haul services.

    So for example, replace all the long haul trucks with driverless trucks that have a fleet of drones in the roof.

    Then the long haul truck loads up in say San Diego and begins driving to LA. The truck pulls over in each small town along the way and then drones fly out of the roof delivering the packages for that town. The drones then return to the truck and the truck continues to the next small town with the drones charging on route.

    This eliminates the long haul warehousing requirements, it eliminates the long haul drivers and then it eliminates all the courier drivers in each town.

    The economics of this is far more feasible than trying to do drone deliveries from a fixed warehouse.
  • Gosh, Ernst, tell us how you really feel!  :)

    Count me in the nay-sayer's column for general-duty consumer drone deliveries.  I can't see how this will make any practical or economic sense anytime soon.

    Joseph, how would solar panels solve any problems at all?

  • The first joke here is in the picture: DHL is arguably the single worst, most inefficient, near-fraudulent "delivery" service on the planet, at the rear of a pack of similar hustlers. Among other things, if you live anywhere at all off the beaten path they will warehouse your shipment in an inaccessible location until they have enough items going to your area to hire someone right off the street, with a private vehicle and no uniform or dress code (like Uber X), and give this person your address and your package. Thus, if this probably otherwise unemployable and sketchy character can actually find you, you'll have the pleasure of being approached by an unidentifiable stranger bearing something you may no longer even need.

    But more to the point,  all drone delivery schemes other than highly specialized one-time fantasies involving heroic dashes with snake-bite antivenom to save Lassie and other such comic-book fodder are, as the realistic Mr. McCray says,  utter nonsense and are nothing whatsoever beyond publicity stunts and click-bait, and always will be.

    One of the larger gorillas in the room is that no matter how advanced the aircraft get, we will continue to live on a planet covered in dirt. A multicopter of useful weight-carrying capacity operating over virgin pavement blows a miasma of filth and dirt over everything and everybody nearby. Perhaps this has escaped the notice of 1%'rs like Bozo the Amazon, who live in polished bubbles with marble floors and whose drone-development minions no doubt wear white lab coats while presenting, with manicured hands, their latest offering to their clueless boss, but out here the world is literally gritty. Perhaps the homeless could be employed to precede the landing aircraft with brooms to sweep the walkway in front of Granny's front door, like a curling team sweeping the ice. Because otherwise the neighbor's Harley is going to get sandblasted, and that will be the end of that particular drone, for sure. 

  • Bah - Humbug!

    "Good analysis" Indeed - NOT!

    Aside from the huge regulatory and insurance hurdles I think the 88 cents per delivery estimate has no relation to reality whatsoever.

    Among other things it would require the support of a huge multi warehouse localized distribution network to have any chance of achieving the proximity necessary, and these would probably only be practical for high population density areas (cities) which have totally problematic structure and traffic density (much higher than normal hazards).

    And the cost of maintaining and operating this service would be astronomical.

    And all that is only on top of the intrinsic safety, regulatory and reliability problems.

    Basically, given anything remotely like the restrictions and limitations of the current technology this is complete Bollocks!

    There are a thousand good things to do with our UAS and at this point in time this is NOT one of them.

  • Admin


    Well stated and I am in complete agreement with you.



  • I envisioned that these drones would be solar powered or augmented by solar which would solve part of the problem
  • Developer

    One advantage I've heard of very fast delivery (i.e. within an hour of the customer's order being placed) is that it greatly reduces the chance that the customer will not be home meaning repeat trips (which double or triple the delivery cost for the package) are greatly reduced.

  • The hurdles are many and large, of course.

    However, on a more optimistic note, given the very deep pockets of the players now trying to do this, it must represent one of the most promising avenues for small startups who can pioneer even just a single innovation that might mitigate, or even plausibly appear to mitigate, any of the technical hurdles.

    Then patent, then wait/hope to be bought out by the big guys.

    This happens a lot in the tech world, even if it leads nowhere much in the end. Drones shouldn't be any exception.

    Is there any other drone application where the players compare in size to Amazon + Google + Wallmart + DHL etc.?

    I'm not saying it will be done (though who would bet against it in the long run?). Just that a lot of dollars will probably change hands in the attempt.


  • Admin

    Once enough delivery drones operating in urban environments crash from either mechanical or control failures, liability insurance will become sky high unless the operators plan to be self insured.



  • Moderator

    I've got some bad news for Tasha Keeney we are many many years out from one pilot multiple RPA. Regulators will just not permit it. Those 6000 pilots would only fly 6000 RPA so the numbers are well out. 

This reply was deleted.