The best coffee I made last winter was an accident born of two disappointments. I had a washed Ethiopian that was all jasmine and lemon and had no middle to it whatsoever, and a Brazilian that was sweet and nutty and about as exciting as a beige wall. Both were roasted, both were fine, neither was a cup I wanted twice. On a morning when I was low on each, I weighed them into the same brew out of pure laziness, and the result was better than either bag had any right to be. The Brazil gave the Ethiopian a floor to stand on. That accident is the whole argument for blending, and it took me another month to learn how to repeat it on purpose.
What a blend is actually for
Blending has a slightly commercial reputation among home roasters, as though a single origin is the honest choice and a blend is what you do when you have something to hide. In a roastery, sure, part of the point is holding a flagship product steady as harvests come and go. At home the reasons are simpler and more selfish.
The first is filling a hole. Plenty of coffees are outstanding at one thing and thin at everything else, and a component that contributes body and sweetness costs you nothing in brightness while making the bright coffee feel complete. The second is using up what you have — two half-empty bags of green that don't justify a full roast on their own. The third is that blending is genuinely interesting, and it taught me more about what each origin contributes than drinking them side by side ever did.
Roast together or roast apart
There are two ways to combine coffees, and they are not equivalent. Green blending means mixing the raw beans and roasting the mixture as one batch. Post-roast blending means roasting each coffee separately, to whatever level suits it, then combining the roasted beans.
Mill City Roasters, who build roasters and teach with them, lay out the tradeoff clearly in their green blending basics piece: in a pre-blend all of the coffees spend the same amount of time in the roaster, and differences in bean density, moisture, and size will create some variety in how each one develops. They pre-blend for efficiency and consider those small variances a feature that adds complexity, while making an exception when beans differ significantly in size or when delicate natural-process coffees need their own heat treatment.
At the scale most of us work, that exception is closer to the rule. A commercial drum holds enough beans to average out the differences; a popper holding a hundred grams does not, and one dense, high-grown bean sitting next to a soft, low-density natural will finish in visibly different places in the same eight minutes. Roasting the parts separately also lets each land where it deserves to, which matters when the components want different levels in the first place — the walkthrough on reading a roast level covers what those differences do to a cup. I roast apart and combine after, and I have never regretted the extra ten minutes.
Three components, and why I stopped at three
Mill City's answer on how many coffees belong in a blend is worth borrowing whole. They note that the vast majority of blends have three components, calling three a safe number, and they explicitly advise against using more than four, because too many parts make an individual dose inconsistent — the smaller the serving, the more likely it is to under-represent whatever makes up a thin slice of the mix.
That dosing argument is even sharper at home, where I am scooping around twenty grams at a time out of a jar. A component sitting at five percent of a blend is roughly a gram in that dose, which is a handful of beans, and a handful of beans is inside the noise of how a jar settles. If a part is too small to survive the scoop, it is not in the blend, it is in the jar.
Giving each part a job
The structure I keep coming back to assigns each component a role before I taste anything, so the tasting is checking a plan rather than groping around.
- The base, around 60 percent. Sweetness, body, and low-risk flavor. A washed Brazil or a mid-grown Central American does this well and is forgiving to roast, which matters because it is most of the cup.
- The accent, around 30 percent. The reason anyone notices the coffee. An East African washed lot, a bright Colombian, anything floral, citric, or berry-ish that would be too much on its own.
- The seasoning, around 10 percent. One clear idea — a natural-process coffee for fruit, an Indonesian for earthiness, or the same base origin taken a shade darker for cocoa and depth.
Sixty, thirty, and ten of a three-hundred-gram jar is a hundred and eighty, ninety, and thirty grams, all weighed on the same scale I use for everything else. Adjust in ten-point moves, not five, because five is inside the margin of how differently two coffees can taste week to week.
Tasting a blend without fooling yourself
The failure mode here is deciding a blend is good because you built it. The fix is a tasting order strict enough to catch you out. Cup the components on their own first, then the blend, all at the same ratio and grind and water — the routine in cupping your home roasts is exactly the tool for this, and it is more honest than brewing a mug and forming an impression.
The question is not whether the blend is nice. It is whether the blend is better than its best component, because if it isn't, you have diluted a good coffee with a worse one and dressed it up as a recipe. My Ethiopian-and-Brazil accident passed that test easily; a later attempt that put two bright African coffees together did not, and cupping them side by side made that obvious in about a minute. Hold the ratio steady and hold your brew numbers steady while you judge, since a shifting coffee-to-water ratio will produce differences you then blame on the blend.
What blending will not do
It will not rescue a bad component. A scorched roast, a stale lot, or a coffee with a defect does not get quieter when you surround it with better beans; it gets spread across the whole jar, and you have now lost two coffees instead of one. Anything going into a blend has to be a cup you would drink by itself, even if it is a cup you find boring by itself.
It also will not make a blend taste the same forever. The lot you bought is finite, and next season's Brazil will not be this season's, so a recipe expressed as percentages and roles survives while a recipe expressed as two specific bag names does not. That is the practical reason to write down what each part was doing and not only what it was.
The card taped inside my cabinet
Each blend gets four lines on an index card: the origin and processing of each component, the roast level and the weight-loss percentage each one hit, the ratio, and one honest sentence about the cup. That last line is the one I actually read later. "Floor is right, accent buried at 20, push to 30" is instruction I can act on next month; "quite good" is not.
Start with the accident I stumbled into. Take the brightest coffee you own and the sweetest, boring one, roast each to where it wants to be, and combine them seventy-thirty by weight in a jar. Brew it at the ratio you always use and cup it against both parts. You will know inside one cup whether you have a blend worth keeping, and you will have a number to move next time instead of a feeling to chase.