Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / How Hardware Startups Use Design for Assembly to Cut Costs
Transcript
- Lucas: So there's this moment every hardware startup founder knows — the first time they watch their product being assembled on a real production line, and they realize it takes twice as long and costs twice as much as they modeled on paper. Luna: And usually it's because nobody asked the simple question: 'How is this thing actually going to go together?' Lucas: Exactly. That question is the heart of what's called Design for Assembly — DFA. It's the discipline of designing a product so it can be assembled quickly, reliably, and cheaply. And I think it's probably the single most underrated skill in hardware startups. Luna: Give us a concrete example. What does DFA look like in practice? Lucas: Let's use a smart lock — a company I've been following called Apera, who launched a retrofit smart lock about two years ago. Their first prototype had 47 parts and required 14 assembly steps, including six screws that had to be driven in a specific order. Their contract manufacturer quoted them $11.50 in labor per unit. Luna: That's brutal for a device that probably retails under $100. Lucas: Exactly. So they went back and applied DFA. They replaced the six screws with four snap-fit clips, combined two plastic brackets into one molded part, and moved from a flex cable to a board to board connector that aligns itself. The final design had 34 parts and 9 assembly steps. Labor cost dropped to $7.10 per unit. Luna: That's a 38% reduction — just from eliminating fasteners and reducing part count. Lucas: Right. And that's the core principle of DFA: every part you eliminate is a part that never has to be handled, oriented, or installed. Every screw you remove is a screw that can't be dropped, cross-threaded, or overtightened. Luna: So it's not just about cost — it's about reducing failure modes on the line. Lucas: Absolutely. There's a classic rule from something called the Boothroyd-Dewhurst method — that's the academic framework for DFA. They say the ideal assembly time for a part is about 3 seconds if it's easy to handle and insert. But if you have to rotate the part, use a tool, or align it precisely, that can jump to 12 or 15 seconds. And when you multiply that over tens of thousands of units, the cost compounds fast. Luna: Which is why the smart-lock redesign focused on self-aligning features. The snap fits had chamfered edges so they'd guide themselves into place. Lucas: Exactly. They designed the parts so that gravity and geometry did the work, not the operator's fingers. That's a hallmark of good DFA — you design for what's called 'top-down assembly,' where every part drops in from above and snaps or clicks into position. No flipping, no reaching around the back. Luna: And no screws. Which reminds me — there's a famous example from Dell back in the 2000s. They redesigned their desktop PC chassis to be tool-less. The entire case could be opened and serviced with just your hands. It cut assembly time by 40%. Lucas: Great reference. And the reason it matters for startups specifically is that your first manufacturing run is usually the most expensive. If you haven't optimized for assembly, you're paying that premium on every single unit — and you might not have the volume to negotiate a lower rate. Luna: So when should a startup start thinking about DFA? Before the first prototype? Lucas: Ideally yes, but I'd say at least by the time you have a functional prototype. The trap that a lot of hardware founders fall into is they build a proof of concept using off-the-shelf modules and 3D-printed brackets, and then they try to cram that into a manufactured product. The result is a design that's optimized for tinkering, not for production. Luna: And then they're stuck with high assembly costs and no easy way to redesign without starting over. Lucas: Exactly. That's why the best hardware startups I've seen do what's called 'design for assembly in parallel' — they have one engineer focused on functionality and another focused on producibility from day one. Even if it's the same person, they make sure to spend time on the assembly sequence before the design is locked. Luna: One tool that helps is the DFM/DFA checklist. I know a few contract manufacturers publish them — like Jabil or Flex — where they list the common rules: avoid tight tolerances if not needed, use standard fastener sizes, design for a single orientation. Lucas: Right. And one of the most practical things I've seen is to physically simulate the assembly process. Take your prototype and a stopwatch, and have someone who's never seen the product try to put it together. You'll immediately spot the steps that are confusing or require two hands or need a special tool. Luna: That reminds me — there's a startup I talked to that made a wearable device. They had a tiny flex cable that had to be inserted at an awkward angle. The operator had to use tweezers. In their DFA review, they realized if they just rotated the connector 90 degrees on the PCB, it became a straight push. Saved 8 seconds per unit. Lucas: Eight seconds doesn't sound like much, but at 100,000 units, that's over 222 hours of labor. At $20 an hour, that's $4,400. And it cost nothing to change the PCB layout. Luna: So the ROI on DFA is almost always positive. The question is whether startups have the discipline to do it before they're under production pressure. Lucas: And that's the thing — once you're in production, changes are expensive. You have to pay for new tooling, retrain operators, revalidate the process. So the window to do DFA is really during the design phase, when a file change costs nothing. Luna: If these conversations are useful for what you're building or running, consider supporting the show. It's what keeps us ad-free and focused on the details that matter. You can do that at buy me a coffee dot com slash fexingo. Lucas: Appreciate that, Luna. And it's listeners like that who let us dig into topics like this — the mechanics of actually making things. So back to DFA: I want to talk about one more principle that often gets overlooked, which is designing for the assembly line itself. Luna: What do you mean? Lucas: So a lot of startups design their product as if it's going to be assembled by a single person at a single station. But in reality, most manufacturing lines break the process into multiple stations — Station A installs the battery, Station B mounts the PCB, Station C snaps the housing. And each station has a target cycle time, usually 30 to 60 seconds. Luna: So if one station takes 45 seconds and the next takes 30, you have a bottleneck. Lucas: Exactly. And DFA can help balance the line. For example, if you have a subassembly that takes 50 seconds, you might split it into two stations by adding a temporary fixture that holds the parts between steps. Or you might combine two steps that each take 15 seconds into one station to free up labor. Luna: That's more about process design than product design, but they're deeply connected. Lucas: They are. And the best hardware startups think about that connection early. I've seen founders who actually visited their contract manufacturer's line and mapped out the station layout during the design phase. They asked: 'Where will this part be fed from? How will the operator pick it? Is there a kitting tray that positions all the parts in sequence?' Luna: That kind of thinking can also reduce quality issues. If the operator doesn't have to search for the right part, they're less likely to install the wrong one. Lucas: Right. And there's a great example from a company that makes IoT sensors. They had a design where the antenna was a separate component that had to be soldered by hand. In their DFA review, they switched to a stamped-metal antenna that was integrated into the plastic housing. No soldering, no separate part, no orientation issue. Assembly time dropped from 22 seconds to 4. Luna: And the part cost probably went up a bit, but the labor savings dwarfed that. Lucas: Exactly. That's the tradeoff you have to evaluate. DFA sometimes means spending more on the part itself — a snap-fit housing might require a more complex mold — but if it saves you 10 seconds of labor per unit, the payback period is usually measured in weeks. Luna: So what's the one takeaway you'd give to a hardware startup listening right now? Lucas: Take your current design, count every screw, every cable, every separate part, and ask: 'Does this need to exist, or can it be integrated into another part?' If you can eliminate just 20% of your parts, you'll likely see a 30-40% reduction in assembly labor. And that's profit that drops straight to the bottom line. Luna: And if you're still in the prototype phase, run the assembly simulation with a stopwatch. It might be the most valuable hour you spend. Lucas: Absolutely. Because in hardware, how you make it is just as important as what it does.