Eric Migicovsky, founder of Pebble, walks through the hardware lessons from building a smartwatch. He warns against over-optimizing manufacturing costs too early, which led to choosing an underpowered chip, and describes how a single-supplier display shortage nearly halted production. The key: build an MVP fast and design flexibility into your supply chain.
Chapters
I don't know how many of you guys— is anyone working on a hardware project? Sweet. I'll make a quick pitch on behalf of hardware. It's a lot of fun to do a hardware startup. It's, as I mentioned before, extremely rewarding to be able to wear or use or touch something that you had a hand in making. And I hope that I'm able to describe it in such a way that people don't think that it's this ridiculously hard thing to do, that it is a fairly approachable, thing to work on. There's some tricks of the trade, like try to get to an MVP just as quick as possible. It's the same as software. You want to get to the point where customers are going to pay you for what you're working on. It's not easy to get to that point, but you should try to do that as quick as possible. And yeah, there's a few more things. I'd be happy to chat with anyone who's interested in working in a hardware startup. Thank you. or even someone who's not interested in working on a hardware startup that I could help convince to work on one. So this was about a year after our first prototype went out the door. We worked with a small— we ended up getting into Y Combinator. We had sold several hundred watches, went down to YC, told them that we had already made money, showed them the prototype. They were like, cool. Y Combinator had only funded about 2 or 3 hardware companies up to that date, but they'd never seen someone with an actual shippable product in their interviews, so I think that's why they let us in. Nowadays they let in a lot more hardware companies, but back then I think we were the second or third. This was about 6 months after we were in YC. We worked with a small contract manufacturer in San Jose. We ended up switching away from China and moved it to San Jose because we wanted to be able to evaluate the quality and have a more hands-on manufacturing process. We were making, yeah, a couple thousand, and this was like our medium scale before we did Kickstarter. This was still the same watch, Impulse, though. So that brings us to about January— oh, yeah, a couple things that we did make a mistake on the Impulse days. We made a ton of mistakes. Some of the things that I want to talk about is just pre-optimizing for the cost of goods. So, When you think about doing hardware, everyone thinks of the cost as how much does it cost to actually physically make this widget? In our case, we tried to minimize the cost of the widget ahead of time. We thought that it would be important to keep the cost of goods, the things that go into the widget, as small as possible. The problem is when you try to do that at an early stage, you end up having to spend a lot more money on other things. Like we chose a really inexpensive microcontroller. the actual computer that goes inside the watch. Problem for us is we ended up having to do a hell of a lot of software work to be able to cram all the things we wanted into this tiny little microcontroller. If we had simply said, OK, we're going to pay $2 or $3 more per unit, we would have had so much more computer to play with, and we wouldn't have had to spend weeks or months cramming all this in there. And that was mainly because we thought we were going to go bigger than we were. Too soon. And if we had just aimed a little bit lower, we would have had a much easier time, and we would have been able to move much faster. We also had to deal with a really stupid supply chain problem where the display that we had selected was just not in stock ever. And we ended up having to buy the display like 8 months ahead, and it took a huge amount of work to just get this display. And one of the things that I was thinking is if we could have designed our product to be a little bit more changeable, if we had said, okay, we could use this display or this display, we would have been able to select from 2 different vendors. And that was something in the short term that almost killed the company because we just didn't have any displays to put into the product. And in hindsight, if we had been a bit smarter about the design process, we probably wouldn't have run into that problem.
