Cupping & sensory

Cupping Score

Scores are the currency of specialty green coffee trading and are routinely quoted with more precision than they can support. This page sets out what the number contains, what makes two scores comparable, and how to put one in a contract without creating a dispute.

Last reviewed July 20263 min read

Quick answer

A cupping score is a numerical summary of a sensory evaluation, produced by scoring individual attributes and combining them. It is only comparable between evaluations that used the same protocol, the same roast standard and a calibrated panel. A score from an unnamed protocol by an unnamed cupper carries very little information.

How scoring forms are built

Scoring systems generally work by evaluating a set of named attributes separately and then combining them into a total. The attribute list varies by system, but typically covers fragrance and aroma, flavour, aftertaste, acidity, body, balance, uniformity, cleanliness, sweetness and an overall impression, with deductions for faults.

The important structural point is that the total is a *construction*, not a measurement. Changing the weighting of attributes changes the total for identical sensory perceptions. This is why the protocol has to be named for the number to mean anything.

Scoring systems change

The bodies that publish cupping protocols revise them. Attribute lists, weightings and the way descriptive and affective assessment are separated have all been reworked in recent years. A score quoted without its protocol and version is not a stable figure, and a threshold remembered from an earlier system may no longer correspond to the current one. Always establish which protocol and version applies to the crop being contracted.

What makes two scores comparable

1The same protocol and version. Different forms produce different totals from the same perceptions.
2The same sample roast standard. A darker sample roast will move several attributes.
3A calibrated panel. Individual cuppers drift. Calibration against shared references is what keeps a panel’s numbers meaningful over time.
4Comparable sample freshness. A score on a fresh sample and one on the same coffee six months later are both valid and are not the same number.
5The same water and equipment. Both move perceived acidity and body measurably.

When any of these differ, a difference of a point or two between two scores carries no reliable information. Treat small differences between separately produced scores as noise.

Using a score commercially

Scores are genuinely useful for two things: sorting an offer list into rough tiers, and setting a floor in a contract. They are poor at fine discrimination and poor as a substitute for cupping the sample yourself.

Sensible and unsensible uses of a score
Use	Assessment
Filtering a long offer list to a shortlist	Sensible — that is what a summary figure is for
Setting a contractual floor with a named protocol and panel	Sensible — provided all the terms are stated
Choosing between two lots one point apart, sight unseen	Unsound — that difference is within noise
Comparing a score from one exporter with one from another	Unsound unless the protocol and calibration are known to match
Replacing your own cupping of the pre-shipment sample	Unsound — the approval must be yours

Writing a score into a contract

What the clause needs

Protocol and version: Named explicitly
Who evaluates: Seller’s panel, buyer’s panel, or an agreed third party
Sample type: Pre-shipment sample, drawn to an agreed method
Score basis: Minimum total, and any attribute-level minima
Fault terms: Explicit — a zero-tolerance clause for taints is usually clearer than a score floor alone
Dispute route: An agreed independent panel, and who pays

A note on how we quote scores

A cupping score quoted against a lot should be a measured figure with the protocol and the number of cups stated. We do not publish estimated or aspirational scores, and we do not quote a score for a lot that has not been cupped. Ask any supplier — including us — which protocol produced a figure before you price against it.

Cup it yourself

The only approval that protects you is your own. Tell us what you are looking for and we will tell you what the current crop supports.

Request Current Crop Offer

Frequently asked questions

What is a good cupping score?

That depends entirely on the protocol, the panel and the intended use. Rather than anchoring on a remembered threshold, establish which protocol and version a score was produced under, whether the panel is calibrated, and then cup the sample yourself against your own reference.

Can I compare scores from two different exporters?

Only if you know both used the same protocol and version and that both panels are calibrated. Otherwise the comparison is not meaningful, and small differences certainly are not.

Why do scores change between the offer sample and arrival?

Because the coffee has aged, travelled and possibly changed moisture, and because a different panel on a different day is a different measurement. Some drift is normal; a large gap points at either sampling or storage and is worth investigating.

Should a contract specify a minimum score?

It can, provided the protocol, the evaluating panel, the sample type and the dispute route are all named. A score floor with none of those attached is very difficult to enforce. A clear fault-tolerance clause is often the more useful protection.

Tell us the coffee you need

Lots can be specified by origin, region, process, grade, screen, moisture, defect tolerance, crop year and packaging. Send what you know and we will confirm what each origin realistically supports.