"Cupping" sounds like something that involves a clipboard, a lab coat, and a certification I don't have. Strip the ceremony off and it's just this: grounds and hot water in two cups, prepared identically, tasted side by side. I roast in small batches and change one thing at a time between them, and until I started doing this I genuinely could not tell you whether any given change had worked. I'd roast a batch on Tuesday, drink it Wednesday, roast another on Saturday, and by then Tuesday's coffee was a vague memory filtered through whatever I'd eaten for breakfast. Cupping is how the roasting log stops being a diary and starts being feedback.
Why side by side beats memory
Two reasons, and they're both about removing noise. The first is that palate memory across days is terrible — far worse than we think it is. You cannot reliably compare a coffee you're drinking now against one you drank four days ago; you can only compare it against your description of it, which was already fuzzy when you wrote it. Put the two coffees on the counter at the same time and differences you'd never have named in isolation become obvious in one alternating sip.
The second is that normal brewing buries roast differences under brewing differences. A pour-over involves pour rate, drawdown time, filter behavior — enough variables that a slightly different cup might be the roast, or might be your wrist. Cupping deletes all of that. No filter, no pouring technique, no device: hot water onto grounds, sitting in an open cup. What's left to vary is the coffee itself, which is the entire point.
The setup
You need two identical cups or small bowls — identical matters more than fancy; matching mugs are fine. Into each goes the same dose: I use 11 grams of coffee to 200 ml of water, which is about two level tablespoons of whole beans if you don't have a scale (get a scale, but that's another entry). Grind both samples on the same setting, medium-coarse, one right after the other, and grind a pinch of the second batch and throw it away first — it clears the first batch's dust out of the grinder so sample B doesn't arrive wearing a little of sample A.
Water is whatever you'd normally brew with, just off the boil — I let the kettle sit for about 30 seconds after it clicks off, which lands it right around 200°F. Beyond that: one timer, one soup spoon, and a glass of plain water for rinsing the spoon. That's the entire equipment list.
The routine, start to slurp
Pour water over the grounds in both cups — enough to wet everything, then fill to your 200 ml mark — and start the timer. The grounds float up and knit together into a crust on the surface. Leave it alone.
At 4:00, break the crust. Push the floating grounds back with the bowl of your spoon, two or three gentle strokes, with your nose right at the rim of the cup. This is the single most aromatic moment of the whole process — the steep's aroma comes up in one concentrated wave as the crust opens — and it's worth doing deliberately, cup A then cup B, because differences often show up in the nose before you taste anything. Rinse the spoon, then skim the foam and any stubborn floaters off both cups and let them sit.
The grounds sink. Somewhere around 9 or 10 minutes in, the coffee is cool enough to actually taste. Take a spoonful and slurp it — genuinely slurp, hard enough to feel slightly ridiculous — then go back and forth between the cups, rinsing the spoon each time. Keep tasting as the cups cool, because the coffee keeps changing: hot coffee hides a lot, and both sweetness and flaws come into focus as the temperature drops. Some of the most useful notes in my log came from cups that were nearly room temperature.
What to write down
This is a log entry, so treat it like one. At the top: the date, the bean, and what was different about each batch — roast date, time to first crack, how many seconds past first crack you pulled it. Then, per cup, three words maximum. Comparative pairs beat vocabulary every time: brighter/flatter, sweeter/harsher, thin/syrupy. And always answer the only question that actually matters: which cup would I rather finish?
Skip the flavor-wheel poetry. "B tastes like A but harsher" tells future-you something actionable; "hints of stone fruit" tells future-you that past-you was feeling fancy. The goal isn't describing coffee well — it's connecting a difference in the cup back to a decision you made during the roast, so the next batch can be a deliberate step instead of another guess.
Cupping against a bag you know
Once the two-cup routine feels comfortable, the most useful third cup isn't another of your own batches — it's a commercial coffee you already like, bought roasted, cupped right alongside your roasts. Nothing about the routine changes; you're just adding a bowl. What it does is anchor your scale. After a few weeks of tasting nothing but your own roasts, "pretty good" starts drifting — every batch is being compared to your last batch, and the whole reference frame can slide without you noticing. A known bag from a roaster you trust resets it.
Fair warning: the first time you do this, it's humbling. A professional roast on a machine with a temperature probe usually beats a popcorn-popper batch, and that's fine — the point isn't to win. The point is to hear what the gap actually consists of. Mine turned out to be evenness more than roast level, which sent me back to smaller batch loads rather than a different pull time. That's a conclusion I'd never have reached tasting my own coffee in a vacuum.
When the cups argue with the plan
The batch you pushed darker tastes ashy and bitter instead of richer: you overshot the development you wanted — pull the next batch sooner after first crack and see if the roast character stays without the char. The lighter batch tastes grassy and sour with a pale, underdeveloped look: it needed more time, not less; that's the underdeveloped end the roast-level guide warns about, showing up in liquid form.
Both cups tasting flat and papery, regardless of roast level, usually isn't a roasting problem at all — it points at stale roasted beans or tired green stock, and no roast adjustment will cup its way out of that. And if the two cups taste identical despite your careful one-variable change, believe the cups: the change was smaller than a home palate can detect. That's not failure, it's calibration — make a bigger, bolder change on the next batch and cup again. Fifteen extra seconds past first crack might be invisible; forty-five won't be.
That's the loop closed. Reading the roast tells you what you were aiming for; ten minutes with two cups and a spoon tells you whether you hit it. One deliberate change per batch, one honest side-by-side, and the log starts compounding.