By the end: Separate description, preference and conformity; turn observations into testable questions; recognize confounded explanations.
Define Good Before Measuring It
Quality starts with knowing what you want to achieve. A specification is a stated requirement; conformity means meeting it. Does the whiskey meet that requirement, match the intended style, or show an unwanted characteristic? Do I like it? These are related but different questions.
Quality framework distinguishing preference, specification, conformity, safety and legal compliance in whiskey production. Four questions about quality. Liking does not establish conformity, safety or compliance. Source and further reading: Strickland — S1, pp. 257–267, 269–272; ch. 23 closing page (printed number illegible) and 319–338; S2, pp. 82–83.Describe Before Explaining
Record sample identity, preparation, conditions, descriptor, intensity and repeatability. “Unexpected aroma compared with the retained reference” is more useful than immediately naming a molecule or a faulty machine.
Analysis answers defined questions about composition; sensory work addresses perception. A pleasing aroma or one laboratory result cannot certify every dimension of quality or safety. Strickland — S1, pp. 257–267, 269–272; ch. 23 closing page (printed number illegible) and 319–338; Evidence note — S2, pp. 82–83.
Trace Back Through The Process
Ask what changed in grain, water, mash, fermentation, distillation, cask, storage, blending or packaging. Records narrow the possibilities; missing records leave guesses.
First distinguish a measurement problem from a product difference. Was strength matched? Was the reference stored similarly? Could samples have been mislabeled? Did the difference repeat?
These are Academy teaching questions built from an integrated process approach, not a validated diagnostic algorithm.
Troubleshooting diagram tracing a whiskey difference from matched measurements back through lot and production records. Trace a difference back. Match samples and conditions, then use lot records to follow the production history. The lot entries shown are illustrative. Source and further reading: Strickland — S1 integrated framework; Academy synthesis.Test An Explanation
Suppose a grain lot and yeast strain changed together and the spirit changed. Either may contribute, as may their interaction or another difference.
A controlled trial holds relevant conditions as steady as practical while testing a defined change. Repetition and suitable measurements help distinguish an effect from ordinary variation. If production constraints limit the comparison, record them instead of overstating the conclusion.
Controlled-trial diagram for testing one defined production change while holding other whiskey-making conditions steady. One change at a time. The lower pair keeps the grain lot constant while changing yeast. Source and further reading: Evidence note — Academy synthesis; S1 integrated framework.Make The Record Useful
Capture the observation, expected result, lot, relevant process records, competing explanations, test, outcome, decision and remaining uncertainty. An honest “not yet explained” identifies further work more reliably than an unsupported causal story.
Try It: Make A Decision Against A Brief
Fictional brief: a batch must match a specified house profile. A panel prefers sample B, but B does not match that profile. Sample A does. Which fits this release brief, and what should happen to the preference finding?
A fits the stated profile on the information given. Record the preference for B separately; it may inform another product or a revised brief, but it does not erase the mismatch. Here, conformity means meeting the stated requirement—not being everyone's favorite.
Academy practice exercise based on this lesson; fictional examples are not producer data.
Check Your Understanding
A panel prefers B, but B misses the stated specification. Which finding counts?
Both. Preference and conformity differ. The producer can investigate, reconsider the brief or reject B for that product; liking does not erase the specification.
Go deeper: An Original Academy Batch Decision Record Use the following prompts to keep a decision traceable. They expand the recordkeeping idea in Strickland's Appendix C; this is an Academy form, not a reproduction of his sheet.
Check the story with arithmetic. A fictional charge contains 100 L at 20% ABV: approximately 20 L of absolute alcohol. Express every outgoing amount on the same basis: liters of absolute alcohol (LAA), not liters of liquid. Product accounts for 12 LAA, reserved recycle for 6 LAA, and residue plus documented losses for 2 LAA. The total is 20 LAA, so that account closes. If claimed outputs instead total 23 LAA, investigate units, measurements, incoming recycle and inventory changes before claiming a yield improvement.
A closing alcohol balance checks the account's consistency; it does not establish flavor quality. All quantities in this exercise are invented and use a common measurement basis.
S10 — Strickland, Batch Distillation, pp. 196–217, 278.
🐇 Down The Rabbit Hole
What evidence would distinguish a product change from a measurement problem? Read S10, pp. 196–217 and Appendix C, p. 278; compare S2, pp. 82–83, on description and liking. Use the optional batch record to connect observations, competing explanations and a test.
Production Research Companion (internal research reference)
