4 min read
Choosing a hardware design firm is one of the biggest bets a founder makes, and one of the hardest to judge. Two quotes for the same product can differ by ten times. Both firms will say they can do it. Both will show you a portfolio. And you usually won’t find out who was right until boards are on the bench, or worse, until the product is in customers’ hands. By then it’s expensive.
A mistake caught on the bench costs an afternoon and a bit of wire. A mistake caught after the boards are built means a new revision: weeks of fabrication and assembly, sometimes a part you’ll wait months for, a schedule that slips while everyone waits, and a steady supply of apologies from everyone involved, all of them sincere and none of them a board. A mistake caught by your customers means a recall. The nasty ones are the ones that go all the way. A board that fails once in ten thousand operations sounds rare until you learn the product does ten thousand operations before lunch. A board that fails only when it gets hot never fails on the bench at all, for the same reason a roof never leaks while the roofer is standing on it. The bench is a dry day: a temperature-controlled, well-lit room where the board is fed clean power, handled gently, and examined by several people in office clothes, and it works perfectly, with what can only be described as smugness. The field is the weather. It’s a sealed box in the afternoon sun with nobody watching, and there, considerably less smug, it fails. Open the box to see what’s wrong and it cools down and works again, and it works best of all when you’ve driven out in your office clothes with a face that says please, just this once, fail in front of me. Close the lid and walk away and it fails. Nobody finds a leak on a dry day. A good firm puts the board in an oven.
So the first question isn’t which quote is right. It’s which quote is right for you. It’s what you’re actually buying, and the quotes can’t tell you that. Only you can, and it comes down to one question:
What would it cost you if this failed?
Not the product. This, whatever it is. You might be buying a product, or a prototype to prove the idea works, or a demo that has to get through one investor meeting, or a single board inside a system your own team is building. Each one fails at a different price. A prototype that exists to answer a question is allowed to fail, as long as it fails for the right reason: one that fails because the idea doesn’t work has told you something, and one that fails because the board was rushed has told you nothing and cost you a month. A demo that fails on the day costs you the room. A board inside a bigger system fails at the price of the whole system. And a prototype has a way of becoming the product, usually in a hurry, so it’s worth asking now whether this one will.
Count the recalls, the service calls, the downtime, the lost customers, the damage to your reputation, or the meeting, or the month. That number sets how much engineering this job deserves. If it’s small, the cheap quote may be the right one. If it’s large, the cheaper quote is not necessarily the cheaper choice in the long run. It’s priced the way a budget airline prices a flight: the ticket is cheap, and everything that makes the flight survivable is not. The bag, or your daughter’s three suitcases, are extra. The seat, by which they mean a particular seat rather than the general concept of sitting, is extra. The water is extra, and your son remembered he was thirsty just when you boarded. And you learn this at the gate, which is where budget airlines prefer to deliver news. In hardware the bag, the seat and the water are the unplanned third revision, the weeks it takes, and the firm you hire to fix the first one’s work. The gate is the bench.
Match the rigour to the risk
A good firm scales its work to that number, and what you want is for that to be decided deliberately, at the start, with you in the room. Otherwise it gets decided halfway through, when the schedule tightens and skipping a check looks like saving time. Sometimes it is. But it’s a bet placed with your money, and you should be the one placing it, not the calendar.
The number isn’t only about complexity, though. A simple board can carry a very expensive failure if it’s somewhere nobody can get to it. At the far end of that scale there’s no fixing it at all.
Once you have the number, the rest is finding a firm that lives up to it, and you don’t need to be an engineer for that part. You can’t audit a schematic. What you can do is ask a firm a handful of questions and learn most of what you need from how it answers them: whether its rate actually pays for the work, whether it understands what you’re building, whether anyone checks the engineer’s work, and whether you can stand talking to them for the months it’ll take. We’ve written a post on each one, because the answers ran longer than this and got more opinionated as they went. Start with the number. It’s the only part of this you can do from your desk, and the only part nobody can do for you.
LET'S CONNECT
Need a second opinion on a quote?
Send it to us. We'll walk you through what the work involves and whether the quote covers it.