Foundational
Coffee cupping, and what it is for
The trade's standard comparison protocol, why every part of it exists, and which parts a home taster can drop.
In short
Cupping is the coffee trade's way of tasting several coffees against each other. Ground coffee goes into identical cups, hot water is poured on, and after a few minutes tasters work round the table with spoons. It looks like a ritual. It is a set of controls: every part of it exists to stop something other than the coffee from varying.
This guide describes the method from two documents that were read in full, the Specialty Coffee Association's 2024 standards and the cupping protocol they replaced, and says which one each figure belongs to. It then covers what a cupping score is, how the scoring changed, what studies have found about agreement between tasters, and what a person at home can usefully borrow.
What cupping was designed to find out
Sourced
The 2024 standard defines cupping as tasting several cups of each sample, with a separate set of beans ground and brewed for each cup. It is explicit about the purpose: cupping is a technique designed to assess lots of green coffee. The coffee has to be roasted, ground and brewed only because there is no other way to taste it.
That is why the method looks the way it does. Grinding each cup separately means one bad bean spoils one cup and not the whole sample, so tasting five cups shows whether a lot is uniform. Steeping in the cup with no filter and no pouring technique removes the brewer's skill. Identical vessels remove the vessel.
The standard also says what it does not cover. Quick cupping of many samples for quality control is, in its words, outside its scope. It is written for arabica and says it may be adapted to other species. And it separates three things the older practice ran together: how a sample is prepared and tasted, how it is described, and how much the taster likes it. Each has its own standard.
The earlier protocol made a related point that is easy to miss. Sensory testing, it said, is done to find differences between samples, to describe them, or to find preferences, and no one test does all three. Cupping to a score is the third kind.
The figures, and whose they are
Sourced
Two documents, read in full. Where one figure is given both documents agree; where they differ both are shown.
| Variable | Typical range | What moving it does |
|---|---|---|
| Coffee per cup | 8.25 g per 150 mL of cup volume | Both documents. The 2024 standard allows 0.2 g either way and has the beans for each cup weighed separately. |
| Cup | 200 to 350 mL (2024); 207 to 266 mL (2015) | Tempered glass or ceramic, all identical. The 2024 standard gives a top diameter of 75 to 90 mm; the 2015 protocol gave 3 to 3.5 inches. |
| Grind | 70 to 75% passing a 20-mesh sieve | Both documents describe this as slightly coarser than for paper-filter drip. Each cup is ground separately, after a little of the sample has been run through to clear the grinder. |
| Water temperature at pouring | 93 °C, give or take 3 (2024); about 93 °C (2015) | Poured to the rim, because the dose was calculated from the cup's volume to the rim. |
| Water | Calcium hardness 50 to 175 ppm and alkalinity at or near 40 to 70 ppm, both as calcium carbonate; pH 6 to 8; no chlorine (2024) | The standard asks for this "whenever possible". The 2015 protocol instead asked for water that was clean and odour-free, not distilled or softened, ideally with 125 to 175 ppm of dissolved solids and never under 100 or over 250. |
| Crust left unbroken | 3 to 5 minutes | Both documents. It is then broken by stirring about three times. |
| Tasting begins | At about 70 °C (2024); 71 °C (2015) | The 2015 protocol put that at about eight to ten minutes after pouring. The 2024 standard asks for at least three rounds as the cups cool to below body temperature. |
| Cups per sample | Five for scoring; three to five for description | Five is what makes the uniformity count possible. |
| Roast of the sample | Medium, to a stated colour reading | Both documents give a reading of 63 on one maker's roast-colour scale for ground coffee. The 2024 standard says coffee should rest 8 to 24 hours after roasting; the 2015 protocol asked for at least 8 hours' rest and cupping within 24, and for a roast of 8 to 12 minutes. |
| Coffees on a table | At most six (2024) | The standard says the maximum should be six, and fewer when tasters are describing and scoring at once. |
These are the parameters of two protocols written for trade tasters assessing green coffee. They are not recommendations for brewing, and neither document claims they make the coffee taste its best.
The sequence at the table
Sourced
As the 2024 standard sets it out. All the coffees on a table go through each step together.
Smell the dry grounds
As soon as practical after grinding. The standard calls this fragrance, and notes that it is purely a matter of smell.
Pour
Hot water to the rim of every cup. Grounds rise and mat together on the surface into what the standard calls a dome or crust.
Smell the crust, then break it
The undisturbed crust is smelled, left for three to five minutes, and then stirred about three times while the taster smells what is released. Together these are the aroma.
Skim
Floating grounds, foam and oils are lifted off with one or two spoons and thrown away.
Wait, then taste in rounds
Once the cups have cooled to about 70 °C, a spoonful from each is slurped so that it covers the tongue and palate, assessed, and spat out. At least three rounds follow as the coffee cools to below body temperature.
Rinse the spoon, every timenot always done
From the breaking of the crust onwards the spoon is rinsed in hot water before it touches any cup. The standard also describes a variant in which each taster transfers coffee to a personal spoon or glass.
A diagram of the published sequence. The slurp is functional: the standard recommends keeping it because it helps odour reach the nose from the mouth.
What a cupping score is, and how it changed in 2024
Sourced
The score most people have met is the hundred-point score of the earlier protocol. In the 2015 edition a taster scored ten things: fragrance and aroma, flavour, aftertaste, acidity, body, balance, uniformity, clean cup, sweetness and an overall mark, with defects subtracted. Each quality mark ran in quarter points on a scale whose printed part went from 6 to just under 10. The protocol's key called 80 to 84.99 very good, 85 to 89.99 excellent and 90 and above outstanding, and placed anything under 80 below specialty.
The protocol described what that number is. Its purpose, it said, is to record the cupper's perception of quality, drawing on the cupper's previous experience. It also allowed for expectation: a coffee expected to be high in acidity and one expected to be low could receive equally high marks for acidity. The score was never a measurement of what is in the cup.
The 2024 standards say they supersede that protocol and split it in two. A descriptive assessment records what the coffee is like: the intensity of each section on a fifteen-point scale, and ticks against a fixed list of descriptors, with no judgement of good or bad. An affective assessment records the taster's impression of quality on a nine-point scale for eight sections, from fragrance to overall. The standard calls that impression subjective in so many words, and says it may reflect the taster's own preference or a market's.
A score still comes out of the affective form, by a published equation: the eight marks are added, multiplied by 0.65625, and 52.75 is added, with two points taken off for each cup that differs from the others and four for each defective cup, and the result rounded to the nearest quarter point. So nines throughout give 100, fives throughout give 79, and ones throughout give 58: the scale cannot go lower than that before deductions. The form has no line for balance or for clean cup. The standard mentions balance as one of the things the overall mark may take in, and it limits sensory defects in practice to three: potato, mouldy and phenolic. It adds that the statistics needed to interpret such scores with confidence, and the selection and training of tasters, are outside its scope.
The association's own summary places both of these beside two further assessments, of the physical green coffee and of information such as certification and origin. It describes the four together as a fuller picture than a single score. The hundred-point score of the earlier form remains in wide commercial use.
How far tasters agree
Sourced
Two studies read for this page bear on it, and they point the same way with different emphasis.
In one, fifty-six coffee experts scored twelve coffees twice, once by their usual method and once by a method the authors had developed. Each expert brewed the coffees themselves, by cupping, drip or espresso as they preferred. The authors report that agreement on which coffees were better and which worse was very high, and conclude that trained experts can share a concept of quality. They show that agreement as the closeness of the experts' arrows on a statistical map of the scores; the paper gives no single figure for it. They also set out long-standing criticisms of hundred-point scoring: that it mixes intensity, quality and the presence of defects in one number, and that it gives little account of why a coffee scored as it did. Those authors devised and trademarked the alternative method, though the paper declares no conflict of interest, so the criticism comes from an interested party. The agreement finding does not obviously serve that interest, which is a reason to give it some weight.
In the other, forty-six experienced tasters in five countries described eight espressos on three occasions each, by ticking descriptors. They were given the lexicon and the guidelines but no training session. Most chose much the same descriptors each time, and not all did: the paper counts thirty-nine of the forty-six as repeatable, from all ten in one country to five of seven in another. Tasters in different countries used somewhat different words for the same coffees: half of the ten country-to-country comparisons reached the level of similarity the authors treat as acceptable, and half did not.
Neither study tested untrained tasters, and neither compared a score given on one day with a score given on another. What they support is modest: experienced people, using a shared form, mostly agree with themselves, and broadly agree with each other about which coffees in a set are better. That is a statement about rank order within a session. It does not make a score from one table comparable with a score from another, and the standard's own remark about statistics is a caution in the same direction.
What is worth borrowing at home
CoffeeHQ explanation
This is CoffeeHQ's reading of which controls do the work. It is not a protocol and nobody here has run it.
- Keep the controls, drop the form. Identical cups, the same weight of coffee in each, the same grind setting, the same water poured at the same moment. Those are what make two cups comparable. The score sheet is what makes professionals comparable with each other, and one person at a kitchen table has nobody to be compared with.
- Taste coffees that differ. Side-by-side tasting shows differences, so give it one to find. Three or four cups is a manageable table for one person.
- Hide the labels if the question matters. Number the cups underneath and shuffle them, or have someone else pour. What a bag says changes what people report tasting, and the only defence is not knowing.
- Rank and describe; do not score. Put the cups in order of preference and write down why. A number produced alone has no outside meaning, and the standards say as much about their own scores.
- Mind the heat and the spoon. The protocols pour water at around 93 °C and begin tasting at about 70 °C, which is hot enough to scald. Wait until a spoonful is comfortable, and rinse the spoon in hot water before each cup, as the standard does; if anyone else is tasting, it also describes each person spooning coffee into their own glass instead of sipping from the shared cup.
Outside this page: thresholds, training and brewing
CoffeeHQ explanation
It does not say that cupping parameters are good brewing parameters, or that a cupped coffee tastes as it would brewed another way. It does not say what score makes a coffee specialty today: the 80-point line belongs to the superseded form, and the present standards define specialty coffee by distinctive attributes and market value, not by a number. It does not describe the training or examinations of professional tasters, which the standards leave to other documents that were not read.
The appendix of smell references in the descriptive standard is not reproduced here, and the fourth assessment in the association's scheme, of information such as origin and certification, was not read.
What to do next
- Tasting Lab — A worksheet for a side-by-side or blind tasting, and a triangle test for checking whether a difference is detectable at all.
- Coffee flavour explained — What the words on a cupping form mean, and which sense each belongs to.
- Coffee grading and quality — The physical side of assessment, which the cupping score is often confused with.
- Buying and judging green coffee — Where cupping sits in a trade: offer, pre-shipment and arrival samples.
- Coffee Expert level 11: cupping and tasting — The same ground as a lesson, with exercises on blind and comparative tasting.