What to look for when buying from independent roasters: the useful version

Judging a roaster is mostly about separating claims that can be checked from claims that cannot. Some things a bag says correspond to a measurable property. Others are vocabulary.

What to take away

• Roast degree is measurable and increasingly measured, so a vague roast description is a choice.
• Origin, variety and roast degree can all be verified analytically, which means fraud is detectable.
• A roaster who publishes specifics is accepting the risk of being checked.
• Two purchases tell you more than any amount of reading.

Roast degree is not a matter of opinion

Roasters describe roast level loosely because the words are not standardized. The underlying property is not vague at all. Work using near-infrared spectroscopy with chemometrics on washed arabica from Ethiopia and Congo, roasted to industry-validated light, medium and dark degrees, found that NIR spectroscopy discriminated the roast degrees as effectively as gas chromatography did, and that the model could predict volatile compounds across chemical classes associated with roast degree.

So the roast degree of a coffee is a determinable fact, and handheld instruments can approach it. A roaster who says "medium, developed for filter" is describing something real. One who says "our signature roast" has chosen not to. The kitchen version is cruder but real, since bean color maps onto a common curve whatever the profile.

Origin claims can be tested too

Adulteration and origin fraud in coffee are real enough that analytical methods exist for them. A non-targeted liquid chromatography and high-resolution mass spectrometry approach classified arabica and robusta samples by production region, variety and roasting degree, and detected and quantified adulteration, distinguishing coffees even from neighboring countries with similar climates, with classification error rates for variety below 8.3 percent.

You cannot run that in your kitchen. What matters is that specific claims are falsifiable, which is why specificity is a signal. A roaster naming a region and a variety is exposed in a way that one saying "premium blend" is not.

The checklist

On the bag or the site	Weight
Roast date	Essential
Process, named	High
Roast degree in actionable words	High
Region, not just country	Medium
Variety	Medium
Awards and scores	Low
Founder story	None for the cup

Buy twice before deciding

A single bag tells you about one lot. Buy the same coffee twice, four weeks apart, and brew both at your settled recipe. Consistency between the two is the thing a roaster is actually selling, and it is invisible on a first purchase.

If the second bag needs a different grind setting to taste the same, that is normal seasonal variation in green coffee, and it is not a complaint. If it tastes like a different coffee entirely, that is a roasting inconsistency, and it is worth knowing before you commit to a subscription or a standing order. Telling the two apart is a matter of naming the category before you name a cause.

This is also the cleanest way to build a picture of what independence does and does not buy you. Do it with a coffee you already know well, because a familiar reference is what makes a small difference visible, and run the two cups together rather than a month apart, following the ordinary rules for a fair comparison.

Common questions

Do cupping scores mean anything? They mean a trained grader scored a lot under a defined protocol. Useful as a threshold, not as a fine-grained ranking between two good coffees.

Is a roaster's own cafe a good sign? It means their retail coffee is probably fresh and that they get daily feedback. Both help.

Should I buy the cheapest thing they sell? As a first purchase, yes. A roaster's entry level coffee shows their baseline competence far better than the showpiece lot they charge most for.

What about blends from independents? A good blend is built for consistency through the year. That is a real skill, and it is harder than it looks.