Roasting coffee is nothing like using a consumer appliance. If I want to cook something in the oven I preheat it, set the temperature, leave it for the designated time, and voila — food is done. I naively assumed roasting would be the same. I have the beans, I know the flavor notes, I press a button, I take them out at the end.
Wrong.
Here, in no particular order, is what I've learned the hard way.
It's more art than science
The roasters I know who've been doing this for years rely far more on intuitive feel than on rate of rise, development time ratios, or any of the other roasting methodologies I read about. I remember asking a roaster I know for advice on how to approach a new bean. The answer, which I've since learned is basically the answer to every question I ask roasters, was "I don't know, just start roasting and figure it out." There's a ton of practiced, hard-to-explain knowledge accumulated over years of doing it. You can read about it endlessly. At some point you just have to roast a lot of coffee.
Batch size is everything
The amount of coffee you're roasting, called a batch size, matters enormously. A profile that works perfectly at 10 pounds will not work at all at 50 pounds. I learned this the hard way when I accidentally misweighed a batch: it came out 3 pounds lighter than it should have been. I followed my standard roast process anyway. The coffee was undrinkable. I now triple-check my weights before every roast.
Finding the reference roast is hard
For each coffee we have a reference roast, which is a standard profile that we benchmark all future roasts against. One profile per coffee. Except finding that reference roast is a bit like Goldilocks. You do a few roasts, cup them, and see which one is best. What happens if none of them are right? You determine why you think they weren't right and guess a few more times. I'd be lying if I said I nailed it on the first try. Or the second.
Your reference roast isn't as fixed as you think
After all that work, you'd think you can lock it in and move on. Also not true. The reference roast needs to change depending on the weather. On a hot day, beans have slightly less moisture, which means they need less heat to hit the target end temperature. On a cold day, the opposite. The roaster doesn't know what season it is. You have to tell it, essentially, by adjusting on the fly.
The stakes are higher than you'd expect
We roast in 55-pound batches. Each one represents roughly $400-600 worth of inventory. Our quality standards are high, which means a roast that's slightly off can't be used. There's no "close enough." Every time I fire up the roaster, that's the number in the back of my head.
Roasting is humbling in a bunch of ways that I didn't expect. The good news is that every mistake is also a lesson. The bad news is that lessons are expensive.