Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / Why Hardware Startups Are Designing for Manufacturing
Transcript
- Lucas: You spend months perfecting a prototype. It works beautifully. Then you hand it to a manufacturer and they say, 'We can't build this.' Luna: Yeah, that's the moment every hardware founder dreads. And it's surprisingly common. Lucas: Right. And the root cause is almost always the same: you designed for function, not for manufacturing. There's a whole discipline around this called Design for Manufacturing, or DFM. Luna: And I'm guessing this isn't just a nice to have. It can make or break your unit economics. Lucas: Exactly. Let me give you a concrete example. There's a small robotics startup I've been following — they make a collaborative arm for warehouse sorting. Their first prototype used a custom-machined aluminum bracket to hold the motor mount. It worked perfectly. But when they went to production quotes, that one bracket cost $18 per unit. Luna: Ouch. For a single bracket? That adds up fast. Lucas: It does. So they went back and redesigned it as a stamped steel part. Same function. Same strength. Cost dropped to $2.40 per unit. And they cut weight by 15 percent. Luna: That's a huge swing. What was the change exactly? Just the manufacturing process? Lucas: Mostly. The machined part had complex internal features because the original designer wanted to integrate a cable channel. The stamped version used a separate clip-on cable guide. Slightly more assembly steps, but the total cost was a fraction. That's DFM in a nutshell: you optimize for how it's made, not just how it works. Luna: If today was actually useful to you, the way these stay ad-free is listener support — buy me a coffee dot com slash fexingo. Lucas: Yeah, it's a small gesture that keeps the show independent. So back to that bracket — the savings let them hit their target bill of materials, which is everything for a hardware startup at scale. Luna: So what are the practical rules? If I'm a founder about to design my first product, what should I be thinking about from day one? Lucas: Great question. There's a classic set of seven DFM principles. First: minimize the number of parts. Every part is a cost, a supplier relationship, a potential failure point. Luna: Right. So combining functions into one part where possible. Lucas: Exactly. Second: standardize components. Use off-the-shelf screws, nuts, connectors. Custom parts are expensive and slow. Third: design for easy assembly — make parts self-aligning, use snap-fits instead of fasteners where you can. Luna: I've seen that in consumer electronics. Those little plastic clips that hold things together. Lucas: Exactly. Fourth: design for the specific process you'll use. Injection molding has different rules than stamping or casting. Fifth: avoid tight tolerances unless absolutely necessary. Every extra decimal place on a dimension doubles inspection cost. Luna: So don't spec a plus or minus 0.01 millimeter if 0.1 works. Lucas: Right. Sixth: design for the material you're using. And seventh: consider the full assembly sequence. If a worker has to flip the part three times, you're adding labor time. Luna: That makes sense. But there's a tension here, right? If you optimize too early for manufacturing, you might miss a better functional design. Lucas: That's the classic counterargument. And it's valid. You don't want to constrain creativity before you've validated the core function. The key is knowing when to switch modes. Luna: So prototype for performance, then redesign for production. Lucas: Exactly. That's the phrase I'd use. Get your first prototype working, prove the concept, then do a DFM pass before you commit to tooling. The bracket example — they didn't change the motor or the arm geometry. They changed how the bracket was made. Luna: And that saved 87 percent on that one part. But is it always that dramatic? Lucas: Not always. Sometimes DFM saves 10 percent. But even 10 percent on a $50 BOM is huge margin. And there's another benefit: reliability. Parts that are easier to manufacture tend to have fewer defects. Luna: Makes sense. So how early should a founder involve a contract manufacturer in this process? Lucas: As early as possible. Ideally before you finalize the design. Most good contract manufacturers have DFM engineers who will review your files and flag issues. Some will even suggest alternatives. Luna: But isn't that risky? You're sharing your design before you have an NDA. Lucas: It's a legitimate concern. But most reputable manufacturers are used to it. You can do a high-level DFM review without revealing proprietary details. You can say, 'I need a bracket that holds this load with this envelope,' and they'll tell you the best process. Luna: So it's more about the process than the exact geometry. Lucas: Exactly. And that's a mindset shift for many engineers. We're trained to design the perfect part. But manufacturing is about trade-offs. Luna: Can you give an example from a bigger company? Something we'd recognize. Lucas: Sure. Think about the Ford Model T. Henry Ford didn't invent the automobile. He invented the assembly line — which is the ultimate DFM innovation. He redesigned the car so it could be built quickly by unskilled workers. That's why the price dropped from $850 in 1908 to $260 by 1925. Luna: And the design itself was simpler. Fewer parts, easier assembly. Lucas: Right. The Model T had a planetary transmission that was notoriously hard to shift. But it was cheaper to make than a standard sliding-gear transmission. Ford optimized for manufacturing cost, not driver comfort. Luna: That's a pretty extreme trade-off. But it worked because the market valued affordability. Lucas: Exactly. And that's the lesson for hardware startups: know which trade-offs your customer will accept. If you're selling to industrial buyers who care about reliability above all, maybe you don't skimp on the bracket. But if you're competing on price, every dollar counts. Luna: So DFM isn't just about making things cheaper. It's about making the right things cheaper. Lucas: Beautifully put. And that's why the best hardware founders think about manufacturing from day one, even if they don't start designing for it until later. Luna: I want to push back a little. There's a risk that DFM can kill innovation, especially in early-stage hardware. If you're too focused on manufacturability, you might never try that wild design that could be a breakthrough. Lucas: That's a real risk. And I've seen it happen. A founder gets a DFM review and the manufacturer says 'we can't do that,' so they water down the design. But the right approach is to ask 'how could we do this?' rather than accepting 'no.' Luna: So find a manufacturer who's a partner, not just a vendor. Lucas: Exactly. A good contract manufacturer will say, 'We can't stamp that complex shape, but we can cast it. Let's talk about the design changes needed.' They help you find the path. Luna: So the bracket example — they went from machining to stamping. That's a process change. But what about material changes? Like switching from aluminum to plastic? Lucas: That's another DFM lever. And it's huge. A plastic injection-molded part can cost pennies, while a machined aluminum part costs dollars. But you need to ensure the plastic meets your strength, temperature, and durability requirements. Luna: And that's where early testing matters. You don't want to discover that your plastic bracket melts at 80 degrees Celsius after you've ordered ten thousand. Lucas: Right. That's why the DFM process includes prototyping the actual production process. You don't just design for injection molding — you run a mold test with the actual material. Luna: Let's talk about tolerance again. You mentioned it drives inspection cost. Can you put a number on that? Lucas: Sure. A typical machined part with plus or minus 0.1 millimeter tolerance might cost $10 to inspect. Tighten that to plus or minus 0.01 millimeter, and inspection can cost $50 or more. Because you need specialized equipment, more time, and often a separate quality engineer. Luna: So every tight tolerance is a cost decision. Lucas: Exactly. Ask yourself: does this dimension actually need to be that precise? Often the answer is no. The part will function fine with a looser tolerance. Luna: What about assembly? You mentioned minimizing parts and designing for easy assembly. Is there a rule of thumb for how many parts is too many? Lucas: Not a hard number, but there's a useful heuristic: for each part, ask if it can be eliminated, combined with another, or made easier to assemble. If you can reduce your part count by 30 percent, you'll typically see a similar reduction in assembly time and cost. Luna: And that's before you even consider the supply chain complexity of managing more parts. Lucas: Right. Every additional part means another supplier, another purchase order, another inventory line. It's exponential in terms of management overhead. Luna: So DFM isn't just engineering — it's also operations and procurement. Lucas: Exactly. And that's why the best hardware startups have cross-functional DFM reviews. Engineering, manufacturing, procurement, and quality all sit in the same room. Luna: That sounds like a meeting I'd want to attend. So what's one actionable takeaway for a founder listening right now? Lucas: Before you send your design to a manufacturer, do a simple DFM checklist. Count your parts. Identify every custom component. Note every tight tolerance. Then ask: can any of these be simplified, standardized, or eliminated? You'll be surprised what you find. Luna: And if you don't have a DFM engineer on staff? Lucas: Hire a consultant for a day. It'll cost a few thousand dollars, but it could save you tens of thousands in tooling and rework. Or partner with a contract manufacturer who offers DFM as a service. Luna: Right. Because a design that's beautiful in CAD but impossible to build is just an expensive picture. Lucas: Exactly. And the best hardware founders know that the real art is designing something that's both functional and manufacturable. That's where the magic happens.