What to write in a roast log so it's useful later
Most roasters start a log, keep it for a few weeks, then stop. It's not laziness: the log asks for effort now and pays back in a month, and by then you've forgotten why you wanted it. The fix is to write down fewer things and make sure each one answers a question you'll actually have.
The question you'll have
It's nearly always the same. "This batch tastes different from last week's. What changed?" To answer it you need two roasts of the same coffee side by side, and enough in each to spot the difference.
What to record every time
What went in. The coffee (and its lot), the charge weight, and the date. The weight matters more than people expect, because it changes how the whole roast runs.
What you did. The charge temperature, and any changes to gas or airflow with the time you made them. If you don't write the changes down, you can't tell your own choices from the machine's behaviour.
The landmarks. The end of drying if you note it, the start of first crack, and the drop, each with a time and a temperature.
What came out. The weight after roasting, so you can work out weight loss.
How it tasted. A line or two after cupping, a few days later. This is the field people skip and it's the one that makes everything else useful. See the guide on cupping notes.
What you can leave out
Anything you can't act on. Barometric pressure, the colour of your apron, a dozen probe readings every thirty seconds you never look at. More data isn't better data. If you can't say how you'd use a field, drop it.
How to read it
Once a month, pick one coffee you roasted several times. Sort the batches by how they tasted, best to worst. Then look down the columns:
- Was the best one longer or shorter than the others?
- Was the development time different?
- Was the charge weight different?
- Did the weight loss land in a different place?
Patterns show up quickly. That's the point of the log: not to record the past but to tell you what to try tomorrow.
Keep it honest
Write the numbers you saw, not the ones you meant to hit. If you planned to drop at 210 and dropped at 214, write 214. A log that holds the plan instead of the reality is worse than none, because you'll trust it.
And write it at the machine, while it's happening. Notes written from memory at the end of the day are tidy, and wrong.