How to Evaluate Matcha Samples Before You Buy
At a glance
To evaluate matcha samples, run every sample through the same fixed sequence: a dry check on one consistent dark background under one light, aroma off the open bag, a sifted whisk test at your own bar's water temperature, then the latte you actually sell. Keep the matcha you serve now in the lineup as a reference, hide the labels, pour in random order, and have more than one person score each attribute on a scale you never change between sessions. Dry colour is the most over-trusted signal: it proves the leaf was shaded, but it also moves with grind fineness and storage age, and it predicts nothing about how the tea holds up in milk. Scored against a reference by a panel, the result is a measurement rather than an impression.
- Dry colour proves shading, not quality. It also shifts with grind fineness, storage age, and added colouring, and Japan's national tea competition scores tencha aroma and taste at 65 points each against 40 for appearance.
- Control everything except the matcha: same water and temperature, same dark background under one fixed light, labels hidden, poured in random order, and always alongside the matcha you serve now.
- The bowl tells you what the tea is; the latte tells you whether you can sell it. Run both, and run iced too if iced outsells hot, before you commit to a lot.
- A bench test cannot detect pesticide residue, heavy metals, or microbial counts. That is the lab's job and it arrives on the lot-matched certificate of analysis.
Most cafe buyers already have a method: smear a little on white paper, read the green, whisk one bowl, decide. It tells you which sample you liked, not why, and not whether you would rank it the same next month. The fix is to hold everything constant except the matcha, and to keep the matcha you pour now in the lineup as the reference.
How to evaluate matcha samples: set the bench first
The matcha is the variable; everything else is a control — and the controls are what most tastings skip.
- Score new samples against the matcha you serve now, not against memory.
- Number the cups and hide the labels. Whoever pours does not score.
- Same water, temperature, volume, whisk, and bowl for every sample.
- Same background and the same light for every dry check.
- More than one taster, each scoring alone before anyone compares.
Whisk it, then put it in the drink you sell
Whisk the sample sifted, so clumping is scored as a sieve problem and not a tea problem. The ISO sensory method evaluates the tea liquor for the levels of frothiness, foam colour and taste, and specifies that correctly made matcha remains in a suspended state without agglomeration or sedimentation. A sample that sediments on your bench will do it again in a to-go cup, and grind is usually why. ISO puts the mean particle size of matcha in the range 10 μm to 30 μm with a median diameter of less than 20 μm, and that is a number you can ask a supplier to produce.
Matcha cupping for cafes: score it, don’t just taste it
ISO 4120 specifies the triangle test, a forced-choice procedure for determining whether a perceptible sensory difference exists between samples of two products. Take its shape, not its statistics: labels hidden, random order, one question before anyone argues about better — are these actually different?
At Japan’s national competition the final score is decided by the judges collectively rather than by any single taster. Score each attribute on a fixed scale and keep the sheets, so you catch drift the year a lot of the same name scores lower on umami. What the bench cannot tell you is whether the lot passed for residue, heavy metals, and microbial counts. That is the lab’s job, on the lot-matched certificate of analysis, read alongside the cultivar and harvest on the spec sheet. When the bench, the certificate, and the spec sheet all check out, put the winner through a week of real tickets before you commit to volume; bulk matcha from 1 kg per tea keeps that first order small.
Where Hisa fits
Hisa buys in Japan on foot, across regions rather than through one producer, so the spec travels with the lot.
Tell us what you pour and your weekly volume, and we’ll put the lots worth testing next to your reference.