DoorDash Built The Drone Last. Start With The Boring Part.

At its Dash Forward event on September 30, DoorDash showed off the aircraft behind its new drone delivery business, DoorDash Air. It has six slow-spinning propellers and a winch that lowers packages to the ground.

The drone got the photos. But the most useful line for founders came from how DoorDash says it built the program. According to TechCrunch, the autonomous drone came last.

Before DoorDash worked on the aircraft, it focused on what happens on the ground: loading systems, how food gets handed off from a restaurant kitchen, and packaging. It sized the whole system around its historical order data, including how much orders weigh and how far they usually travel.

"We didn't start with a drone and ask what would fit," said Harrison Shih, who leads DoorDash Air. "We started with what people order and how local businesses actually operate, then built the entire system from there."

That sentence is a product strategy any early founder can borrow, whether you build drones, software, or a service business.

The company has been unusually open about its build order. Here is the sequence, pieced together from its July announcement and the September event.

1. The data. DoorDash says it has more than 10 billion lifetime orders. It found that more than 20% of orders in 2025 traveled three to five miles, and those took nearly 25% longer than shorter trips, largely because finding a Dasher for mid-range trips takes longer. That was the specific gap a drone could fill.

2. The ground handoff. DoorDash said the harder problem is not the aircraft. It is real-time inventory checks and handoff systems for drive-throughs, rooftops, and restaurant back doors.

3. The routing software. DoorDash Labs, the team that built the Dot sidewalk robot, also built the Autonomous Delivery Platform. It decides which orders make sense to send by air based on weather, range, and ground traffic.

4. The permission. In July, DoorDash received FAA Part 135 air carrier certification, which the FAA uses for drone operators carrying packages for pay. DoorDash says it is the eighth drone operator to earn it.

5. The aircraft. Last.

Founders love the visible part of a product. It is fun to build and easy to demo. Investors like it in a deck. Press covers it.

But the visible part is rarely what decides whether customers keep using you. Delivery is a good example. A drone that flies well but cannot pick up food from a busy kitchen, or arrives with soup spilled, is a failed product. The customer never thinks about propellers. They think about whether dinner arrived hot.

Most products work the same way. For a scheduling app, the hard part is syncing with messy calendars. For a marketplace, it is trust and payments. For an AI tool, it is getting clean data in and putting useful output somewhere the user already works.

When you build the visible part first, you design it around guesses. When you build the plumbing first, you design the visible part around facts.

You do not need 13 years of order data to use this approach. You need to answer four questions before you build the part people will see.

1. What do customers actually do today? Not what you hope they will do. Watch five of them complete the task your product will replace. Write down every step.
2. Where does the work break down? DoorDash found a slow spot in three-to-five-mile trips. Find your version: the step that takes longest, costs most, or fails most often.
3. What has to be true at the handoffs? Every product touches something it does not control: another app, a person, a physical place. List those touch points and test each one.
4. What constraints should size the product? DoorDash sized the aircraft by order weight and distance. Your constraints might be file sizes, budgets, response time, or how long a user will wait.

Only after you answer these should you design the polished part.

If you want help writing these answers down as a simple plan, a tool like Foundra can walk you through mapping the customer workflow before you commit to a build. A notebook and a few customer calls work fine too.

DoorDash's real advantage, as TechCrunch noted, is that it is not a hardware company. It is a company sitting on years of data from merchants and customers. It turned that into a design brief for the drone.

Early startups have data too, just less of it. Some places to look:

• Support tickets and emails. What do people ask for help with most?
• Your own usage logs. Where do users drop off? What do they do over and over?
• Sales call notes. Which objections come up every time?
• A manual pilot. Before you automate anything, do it by hand for ten customers. You will learn the real weights and distances of your own problem.

The goal is to let real behavior pick the first version of the product, instead of picking it in a brainstorm.

Notice that DoorDash did not wait for drones to use what it built. The routing platform already coordinates human Dashers, Dot robots, and outside autonomous partners. The ground work had value before anything flew.

That is a good pattern for small teams. If you build the plumbing first, try to make it useful on its own.

A team building an AI sales agent might first ship a clean way to pull CRM data into one view. A team building a hardware device might first ship the companion app that tracks the problem by hand. A team building a marketplace might first run matching manually over email.

Each of these earns revenue or learning while the bigger product is still being built. DoorDash says some locations in its earlier drone partner pilots saw order volume grow by roughly 30%, a number it could only measure because the ground systems were already running. It also proves the plumbing works with real customers before you add the expensive layer on top.

Systems-first is not free of risk.

You can build plumbing forever. Infrastructure work can expand to fill any amount of time. Set a date when the visible product must ship, even if rough.

You can lose the story. Investors and early users need something to get excited about. Show the vision in sketches or a demo while you build the foundation.

You can over-fit to old data. Historical behavior shows what people did with yesterday's options. A faster delivery mode may create new demand that your data cannot see. Leave room to learn.

Permissions take time. DoorDash needed federal certification before commercial flights, and the FAA notes that operators may still need airspace approvals. If your product touches a regulated area, start that clock early.

1. Write down the visible part of your product in one sentence.
2. List every step that has to happen before and after that part does its job.
3. Circle the step you know least about. That is your first build.
4. Pull one real number from your data or a manual test that should size the product.
5. Decide what the plumbing can do alone so it earns its keep while you build the rest.

DoorDash's pilot deliveries are starting in Northern California with partners including Chipotle and Popeyes. Whether drones become a big part of delivery is still an open question. But the order of operations is a lesson that holds either way.

What is DoorDash Air? DoorDash Air is DoorDash's in-house drone delivery program. It received FAA Part 135 air carrier certification in July 2026 and showed its six-propeller aircraft on September 30.

Why did DoorDash build the drone last? The company says it focused first on ground infrastructure such as loading, kitchen handoffs, and packaging, and sized the system around its historical order data.

What is FAA Part 135 certification? It is the certification the FAA uses for operators carrying packages for pay by drone. It is an important step, but operators may still need airspace approvals.

What does systems-first product strategy mean? It means building and testing the workflows, data, and handoffs around a product before building its most visible feature.

Does this apply to software startups? Yes. Integrations, data quality, and handoffs to other tools often decide whether software sticks, more than the main screen does.

1. DoorDash's drone strategy started on the ground, TechCrunch, September 30, 2026
2. DoorDash Is Cleared for Takeoff: Launches DoorDash Air, Our In-House Drone Delivery Program, DoorDash, July 29, 2026
3. DoorDash is building its own drone delivery business, TechCrunch, July 29, 2026
4. DoorDash unveils Dot, its autonomous robot built to deliver your food, TechCrunch, September 30, 2025
5. Package Delivery by Drone (Part 135), Federal Aviation Administration