Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / How Hardware Startups Use Failure Mode Analysis to Avoid Recalls
Transcript
- Lucas: If you're building a hardware startup, there's one document that separates the pros from the amateurs — and it's not a patent or a pitch deck. It's something called a Failure Mode and Effects Analysis, or FMEA. Luna: FMEA. I've seen that acronym on manufacturing floors, but I think most founders outside of aerospace or automotive have no idea what it actually is. Lucas: Exactly. And that's a blind spot. FMEA is basically a structured worksheet where you list every way a product could fail — not just in the field, but during assembly, shipping, even how the user might misuse it. Then you score each failure mode on three dimensions: severity, occurrence, and detection. Luna: Severity is how bad it is if it happens. Occurrence is how likely. Detection is how easily you'd catch it before it reaches the customer. Multiply those three numbers and you get a Risk Priority Number, or RPN. Higher RPN means you fix it first. Lucas: You've done this before. Luna: I read a lot of engineering textbooks in my spare time. Lucas: Well, then you know the classic example. When Pebble was designing the original smartwatch back in 2012, they ran an FMEA on the battery charging circuit. They identified a failure mode where a faulty charger could overheat the battery. Severity was high — fire risk. Occurrence was low, but detection was also low because the thermal cutoff wasn't redundant. So RPN was high enough that they added a second thermal fuse. Luna: And that saved their launch. I remember reading that a batch of chargers later failed in testing, but the second fuse caught it. Without FMEA, they might have shipped a fire hazard. Lucas: That's the power of the tool. It forces you to think about failure before you're in crisis mode. Most hardware startups skip FMEA because they think it's bureaucratic — something big companies do. But the reality is, the cost of finding a failure mode on paper is essentially zero. Finding it after you've molded the plastic and populated the PCB? That's tens of thousands of dollars. Luna: So where does a startup even start? Do they need a consultant? Is there a template? Lucas: There are free templates everywhere — the AIAG and SAE standards have basic formats you can download. But the key is to involve at least three people: your mechanical engineer, your electrical engineer, and someone who actually builds the thing, like a technician. You sit in a room with a whiteboard and a spreadsheet, and you brainstorm every single failure mode. Luna: That sounds like it could take weeks. Lucas: It can, for a complex product. But for a simple IoT sensor with ten components, you can do a preliminary FMEA in an afternoon. And here's the thing — once you have the document, it becomes a living reference. You update it after every prototype test, every field trial. It's like a risk scorecard that gets more accurate over time. Luna: I've heard that contract manufacturers now ask for FMEA documentation before they even give you a quote. Is that true? Lucas: More and more. Especially for medical devices, automotive, or any product with safety implications. The CM wants to see that you've thought through failure modes because if your product fails, it could damage their reputation too. Some manufacturers even have minimum RPN thresholds — if any risk scores above a certain number, they won't touch it until it's redesigned. Luna: So FMEA becomes a gate to even getting manufactured. That's a big shift from five years ago. Lucas: It is. And it's actually a positive development. Because when you have an FMEA, you also have a much better conversation with your CM. You can say, 'This failure mode has a severity of 9 and an occurrence of 4. Can we add an inspection step to improve detection?' They'll take you seriously because you've done the homework. Luna: Honestly, if today's conversation was worth a coffee to you, there's a simple way to support the show — buy me a coffee dot com slash fexingo. That's all lowercase, fexingo. It keeps this podcast completely ad-free and lets us dig into topics like this without any sponsor influence. Lucas: Yeah, and we really appreciate that. It's the smallest signal that this kind of deep-dive content matters to you. And it directly funds more research time for episodes like this one. Luna: Right. Back to FMEA — I want to talk about a specific case where FMEA saved a startup millions. There's a company called VitalConnect that makes wearable vital-sign monitors. They ran an FMEA and found a failure mode where the adhesive could cause skin irritation after prolonged wear. Severity was moderate, occurrence was high, detection was low. So they redesigned the adhesive and did extensive patch testing before launch. Lucas: And that prevented what would have been a massive recall. I've seen estimates that a recall for a medical device costs around $2 million on average — and that's before legal liability. VitalConnect spent maybe $50,000 on the FMEA and redesign. That's a forty-to-one return on investment. Luna: And it's not just medical. A startup I know in the smart-lock space did FMEA and discovered that a certain type of latch could jam if the door was misaligned by more than three degrees. They added a shim and changed the latch geometry. That cost them two weeks in design but saved thousands of angry customers. Lucas: That's a classic example of the occurrence axis. The door misalignment was common in older buildings — high occurrence. But the detection during factory testing was low because they tested with perfectly aligned doors. The FMEA forced them to test with misaligned doors. That's the whole point: it changes what you test. Luna: One pushback I hear from founders is: 'We're moving fast. We don't have time for a process.' How do you respond to that? Lucas: I say FMEA actually saves time. Because it prevents you from chasing random defects. You know exactly which failure modes to prioritize. And the act of doing the analysis often reveals design simplifications. One hardware founder told me his FMEA showed that a complex mechanical latch had thirteen failure modes. He replaced it with a magnetic catch that had only two. Simplified the product, improved reliability. Luna: So FMEA can actually lead to lower part counts and lower costs. That's a strong argument. Lucas: Absolutely. And there's an emerging trend where startups are using FMEA not just for the product, but for the manufacturing process itself — Process FMEA, or PFMEA. They analyze each step in assembly: soldering, screwing, calibration. What could go wrong at each station? How do you detect it? That reduces yield loss and rework. Luna: That sounds like lean manufacturing principles applied to the startup context. I've seen PFMEA reduce scrap rates by 30 percent in some low-volume electronics assembly lines. Lucas: Exactly. And the beauty is, you can start with a simple spreadsheet. You don't need expensive software. Just a cross-functional team, a whiteboard, and the willingness to imagine the worst-case scenario. Then you fix it before it happens. Luna: Alright, so what's the one takeaway for a hardware founder listening right now? What should they do this week? Lucas: Download an FMEA template — the AIAG one is fine. Gather your two or three key team members. Pick one subsystem of your product — the power supply, the wireless module, whatever. Brainstorm every failure mode you can think of. Score them. Then take the top three RPN items and create a mitigation plan. Do that in one afternoon, and you'll already be ahead of most startups. Luna: And if you're pre-revenue and just prototyping? Is it still worth it? Lucas: Even more so. Because the earlier you catch a design flaw, the cheaper it is to fix. An FMEA done at the prototyping stage can save you from ordering a thousand units of a board that has a fatal flaw. I've seen that happen — a startup ordered two thousand PCBs without an FMEA, and every single one had a ground loop issue. They had to scrap the whole batch. That's a $50,000 mistake that a two-hour FMEA could have prevented. Luna: Ouch. So FMEA is essentially insurance against expensive learning. Lucas: Exactly. And the best part is, once you have a culture of doing FMEA, it becomes part of your design process. You don't even think of it as extra work — it's just how you build things. And your contract manufacturers will love you for it. Luna: And your customers will never know about all the failures you didn't ship. That's the quiet win. Lucas: Right. The best hardware products aren't the ones that never fail — they're the ones where the team thought about failure early enough to prevent it from reaching the user. FMEA is the tool that makes that happen.