Separate three kinds of judgement
Describing a coffee, rating the intensity of an attribute and expressing liking are different tasks. A person can identify a strong fermented-fruit character and enjoy it; another can identify the same character and dislike it. That is not necessarily a failure of calibration. Problems arise when the team mixes description and approval in one unexplained number. Agree which task each field on the form is intended to capture.
The Specialty Coffee Association's Coffee Value Assessment distinguishes types of information that are often collapsed in informal discussion. Use the relevant official materials if that is your chosen framework. For descriptive vocabulary, the World Coffee Research Sensory Lexicon and the SCA Coffee Taster's Flavor Wheel offer valuable starting points. Neither resource proves that every cup must fit a particular descriptor or that every assessor will experience it identically.
Make references reproducible
A word becomes more useful when the team knows what it refers to. Choose appropriate, food-safe references and prepare them consistently. Record the material, preparation, presentation and date. “Like a berry” is a broad association; a documented reference makes the intended meaning easier to discuss. Avoid improvising substances that are unsafe to taste or inhale. Check allergies, dietary restrictions and the suitability of references before a session.
The World Coffee Research lexicon includes reference approaches and intensity concepts that can support training. Consult the source rather than reconstructing recipes from memory. ISO 8586 addresses the selection and training of sensory assessors, while ISO 8589 concerns the design of sensory test rooms. These are useful professional resources, not automatic credentials for a team that has read a summary. Training still needs a defined purpose and a suitable environment.
Get independent notes before discussion
A group can reach agreement quickly by following its most confident member. That is social convergence, not necessarily better sensory evidence. Start with blind codes and quiet individual assessment. Ask people to record what they perceive before anyone names a descriptor aloud. Keep those original notes. Later discussion can refine understanding, but it should not overwrite the evidence of what each assessor initially reported.
Separate the facilitator's role from the desire to produce a tidy consensus. Invite quieter assessors to explain their observations without making them defend a personal preference. When terms differ, explore whether the disagreement concerns the sensation, the vocabulary or the scale. “Bright” and “sharp” might describe a similar perception with different emotional weight. The team needs clearer definitions before it needs a single preferred adjective.
| Observed issue | Possible next step | Avoid |
|---|---|---|
| Different attribute words | Compare safe, documented references | Declaring the senior taster correct by default |
| Different intensity ratings | Review scale anchors and presentation | Assuming liking explains every difference |
| Variable repeat ratings | Use blind repeats and check preparation | Blaming the assessor before checking the setup |
Check consistency, not just agreement
Repeatability concerns how consistent an assessor or process is under defined repeat conditions. Agreement concerns how closely different assessors align. A panel can agree with itself on a shared but poorly defined habit. It can also show useful individual consistency while using different scale ranges. Design exercises that distinguish these situations. Include blind repeats where appropriate and retain the serving order and preparation details.
Do not diagnose performance from one difficult session. Fatigue, illness, adaptation, distractions and preparation variation can affect results. Build breaks into longer sessions and allow assessors to report temporary limitations. For formal performance evaluation, use a suitable design and statistical expertise rather than treating a small collection of scores as a robust ranking of tasters. A training record should help people improve, not merely label them.
Keep preference in its proper place
Preference matters when the question concerns acceptance, a customer group or a product brief. It is not an error to be eliminated from all coffee work. The mistake is presenting preference as if it were an instrument reading or a universal property of the coffee. Explain who made the judgement, under which framework, and for which purpose. A trained panel is not automatically representative of a market.
Close calibration sessions with a concrete change: a clarified term, an updated reference procedure or a repeat exercise. Record unresolved differences instead of forcing a final agreement. Revisit the method when the panel changes, new references are introduced or the coffee set becomes unfamiliar. Calibration is ongoing maintenance of a working language. Its value appears in clearer decisions and more interpretable records, not simply in a narrower spread of scores.