Two of us in a shop, bored, flying a little quadcopter into a garage door at about thirty miles an hour. All four propellers snapped. We went looking for replacements and couldn't find any, not from the manufacturer, not from the hobby shop, not anywhere we thought to look.

So we ordered five hundred sets from China for around five hundred dollars, bagged them one at a time with a label printer, and put them on Amazon. That box turned into roughly a thousand dollars and sold through in a month. Then we did it again. It paid for nachos and beer after work and eventually a kegerator in the office. Not a fortune. But it's the only thing we ever picked that worked on the first try.

Everything we chose on purpose went nowhere. We researched categories, ran the margins, made a plan, bought inventory that sat. One of those experiments left fifteen hundred novelty items in a garage that had to be unpacked, cleaned and repackaged by hand, because the boxes arrived in Chinese. We sold about six of them in a year. My business partner's mother found the rest years later while helping him move, and drew her own conclusions about him.

Some of that is luck and I'm not going to pretend otherwise. But there's a difference I can describe, and it isn't luck. With the propellers, I had the problem in my hands ten minutes before I had the idea. I wasn't estimating whether the annoyance was real. I was annoyed.

Everything else, we were standing outside the market describing it. Demand looked fine on a spreadsheet. The spreadsheet didn't mention the forty other sellers racing each other to the bottom, or the fact that none of us actually cared about the product. That last part sounds soft and it isn't — not caring is why we never noticed the thing that would have told us it wouldn't work.

Here's the part I still think about. My research never surfaced the propellers, and I don't think it could have. The part didn't exist. Nobody was buying it, nobody was reviewing it, nobody was ranking for it. If I'd gone looking for opportunities in that category I'd have found the products that were already there, which is exactly what my research kept finding, in every category I pointed it at.

The sales didn't prove that every person with one of those toys needed propellers. They proved enough people did, which is all a five-hundred-dollar bet requires. But I only knew to place the bet because I was the one holding the broken thing.

I don't have a method to sell you. You can't schedule being the customer, and telling someone to go get interested in something is useless advice. What I have is a rule of thumb I trust more than I used to: when I catch myself annoyed by something I actually use, that's worth more of my attention than a category I chose because the numbers looked reasonable.

The numbers have been reasonable plenty of times. It's the broken propellers that paid for the kegerator.