Coffee is full of confident statements of very different standing. Sorting them is the most useful advanced skill there is, and it is the system CoffeeHQ uses on its own pages.
A measured fact is something somebody measured and reported with a method: the caffeine in a sample, a country's production in a given year. A sourced description is what an authoritative body says: a standard, a definition, a protocol. A common tendency is a generalisation that holds often and has exceptions: naturals are fruitier, altitude brings acidity. A reasoned explanation is a mechanism worked out from what is known, plausible and not directly tested. And a personal note is what one person tasted on one day.
All five are useful. The mistake is promoting one to a higher rung: reading a tendency as a law, an explanation as a measurement, or one person's note as a fact about the coffee.
Five kinds of statement about coffee
CoffeeHQ explanatory model — not a measurement
From least supported (first) to most supported (last).
A personal noteWhat one person tasted or found, once.A review, a tasting note, your own worksheet.
A reasoned explanationA mechanism worked out from what is known, not directly tested.Most explanations of why a technique works.
A common tendencyA generalisation that holds often and has known exceptions.Naturals are fruitier; altitude brings acidity.
A sourced descriptionWhat a standards body or authoritative reference states.A grading protocol, a definition, a standard.
A measured factSomething measured and reported with a method, a date and a unit.Production in a marketing year; caffeine in a sample.
Basis: The distinction CoffeeHQ applies to its own pages, set out as a ladder of support. Limits: The rungs describe how well a statement is supported, not whether it is true. A personal note can be right and a measurement can be wrong.
Coffee science is real and growing, and it is also small, recent and often funded by the industry it studies. A few questions get most of the value out of any paper or news report about one.
What was actually measured — taste by a trained panel, preference by consumers, a chemical concentration, or a computer model? Each supports different conclusions. How many coffees, and which? A finding on one coffee, roast and grinder may not travel. Was it blind? Was there a comparison, and was only one thing changed? Who paid for it, and does the result suit them?
Be especially careful with headlines. 'Scientists find the perfect cup' typically reports a model or a single experiment, and the paper itself is usually far more modest than the article about it. Where you can, read the abstract.
Most questions a drinker cares about — does this gadget, this water, this technique make a difference to me? — can be tested at home well enough to act on. The design is always the same.
Change one thing. Hold everything else as constant as you can, and write down what you could not hold. Taste blind, so that you do not know which cup is which. Use a triangle test to find out whether you can detect the difference at all before asking which you prefer. Repeat, because a single result can be luck. And decide beforehand what result would change what you do.
The last step matters most. If a difference is real and you cannot detect it, it does not matter to you. If you can detect it and do not care, it does not matter either. A great deal of coffee equipment sits in one of those two categories for most people.
Production and trade statistics look authoritative and need the same care. Check the unit: coffee is counted in 60-kilogram bags by most international bodies, and sometimes in tonnes or pounds. Check the period: a marketing year is not a calendar year and differs between countries. Check whether the figure is a forecast, an estimate or a revised final number, since forecasts are routinely revised. And check the source: the main international series do not always agree with each other, because they use different methods.
The same applies to figures about the cup. Caffeine content varies widely with species, dose and method, so any single figure is a midpoint of a range. Extraction percentages depend on how the measurement was made. A number with no unit, date or source is a decoration.
Your task. Pick an accessory you have been tempted by and run it through the checker. Then write down, in one sentence, the experiment that would tell you whether it makes a difference you can taste.
What to notice. The checker answers for a category of setup, not for you. Its confidence label says so. The experiment you wrote is the part that applies to your kitchen.
Exercise: Auditing a coffee claim
Objective. Practise deciding how much a confident claim is worth before acting on it.
You will need
One claim you have recently read or heard — about a gadget, a technique or a health effect
Something to write on
Keep the same
Use the same five questions every time
Procedure
Write the claim as one sentence that could be true or false.
Ask who is making it and whether they sell something that depends on it.
Ask what kind of evidence is offered: a measurement, a blind comparison, one person's taste, or none.
Ask whether the comparison held everything else constant.
Ask how you could test it yourself with a triangle test, and whether the result would change what you do.
Place the claim on the ladder in this level: measured, sourced description, common tendency, reasoned explanation, or one person's note.
What to notice
Did rewriting the claim as one sentence change how strong it sounded?
Is the evidence a measurement of the thing claimed, or of something nearby?
Would a positive result matter in your kitchen?
Reading what you found
Most coffee advice is reasoned explanation or one person's note, and plenty of it is still useful. The point is not to dismiss it but to hold it with the right amount of confidence — and to notice that the same test applies to everything CoffeeHQ publishes.
What this cannot show
An audit tells you how well supported a claim is, not whether it is true. Weakly supported claims are sometimes right.
A protocol for you to run. CoffeeHQ has not run it and reports no result from it.
A reviewer says a new distribution tool 'transformed' their espresso: sweeter, more consistent, higher extraction. They show two shots, one before and one after. The manufacturer supplied the tool.
Reasoning it out
Two shots are two data points; shot-to-shot variation in espresso is large.
The reviewer knew which shot used the tool, so expectation was not controlled.
They had a reason, conscious or not, to find a difference.
'Higher extraction' was not measured, or no method is given.
None of this shows the tool does nothing. It shows the claim has not been tested.
What to do
Place the claim on the ladder: a personal note, presented as a measured fact. To move it up would take repeated shots, blind tasting and a measured yield. Until then the right response is interest without belief — and the option of running the test yourself with a borrowed tool.
UC Davis Coffee Center research — University of California, Davis — Coffee Center
International Coffee Organization trade and production statistics — International Coffee Organization (ICO)
Sections are labelled with the kind of claim they make. How CoffeeHQ uses sources, and what each one is relied on for, is set out in the source and evidence policy.
Kept in this browser only, so you can see where you left off. It is not sent to CoffeeHQ or anyone else, and clearing your site data removes it.