How much matcha per latte: the cafe dose and temperature table
Buyer briefing
What to know
For a 12 oz cup, a hot matcha latte takes 3–4 g of matcha whisked into 30–40 ml of paste before 90–180 ml of milk, and an iced latte takes 4–5 g with 40–60 ml of hot paste over ice and cold milk. A straight matcha shot is 2 g into 30 ml, and a usucha bowl is 2 g into 60–80 ml. For a 16 oz cup, add roughly 1–2 g, which puts a hot latte at 4–6 g and an iced latte at 5–7 g. Hold the water at 70–80 °C (158–176 °F) and steamed milk at 60–65 °C (140–149 °F).
Decision checks
- Dose by cup size and drink type, not by habit: 12 oz hot latte 3–4 g, 12 oz iced 4–5 g, and add roughly 1–2 g for a 16 oz cup.
- Hold water at 70–80 °C. Boiling does not scorch matcha, it over-extracts the bitter catechins, so a harsh cup means check the kettle before you blame the lot.
- Buy a 0.1 g scale and a 60–100 mesh sifter, then pick one point inside each range and write it on the recipe card. A scoop swings 0.5–1.5 g on a 2 g target, and consistency beats the exact number.
Two baristas, one tin, two different drinks. That is almost never the matcha. It is the dose, and it is the one variable on a bar that a written number fixes permanently. How much matcha per latte comes down to two things: the size of the cup, and whether the drink is hot or iced. For a 12 oz cup, a hot matcha latte takes 3–4 g of matcha and an iced one takes 4–5 g, with the water for the paste held at 70–80 °C. Everything below is the full table and the reasoning behind each number, so the number survives contact with a barista who would rather eyeball it.
How much matcha per latte, by cup size and drink
| Drink | Dose | Water and build |
|---|---|---|
| Usucha, the bowl | 2 g | 60–80 ml, all at once |
| Matcha shot | 2 g | 30 ml |
| Matcha latte, hot | 3–4 g | 30–40 ml paste, then 90–180 ml milk |
| Matcha latte, iced | 4–5 g | 40–60 ml hot paste, then ice and cold milk |
Read down the column and the logic shows itself. The bowl is 2 g into 60–80 ml, all at once, because nothing else in the cup is competing with it. The matcha shot is also 2 g, pulled into 30 ml, and that one is the cafe standard, because it drops straight into an espresso-shaped workflow your bar already runs. A hot latte is 3–4 g whisked into 30–40 ml of paste, then 90–180 ml of milk, because milk masks bitterness and mutes flavor at the same time, so a bowl dose vanishes into it. Iced goes to 4–5 g with 40–60 ml of hot paste, then ice and cold milk, since melt dilutes whatever you just built.
Those doses are written for a 12 oz cup. For 16 oz, add roughly 1–2 g. Written out at that size, a 16 oz hot latte lands at 4–6 g and a 16 oz iced latte at 5–7 g. Matcha dose 12 oz vs 16 oz is the line most recipe cards skip, and skipping it is exactly why the large tastes like a small that someone watered down.
The matcha to water ratio a cafe can actually run
The ratio worth controlling is not matcha to finished drink. It is matcha to the small slug of water you wet the powder with first. Grams of matcha per latte only behave if that paste stage is fixed, because that is where the powder either hydrates fully or seals itself into lumps that no amount of milk hides.
So the sequence is: sift, weigh, wet, whisk, build. Add a little of the 70–80 °C water and work it against the wall of the vessel until it is a smooth, glossy paste with no dry pockets, then bring in the rest. Adding matcha to a full cup of liquid does the opposite of what it looks like it should do.
Then aerate. A chasen needs 20–30 seconds of brisk zig-zag from the wrist, not a circular stir. An electric frother gets there in 10–20 seconds and removes the skill variable, which is usually the right trade on a latte-led menu where milk covers the difference in foam texture anyway. Pick one tool per drink type and write down which.
Matcha water temperature, and why “boiling scorches it” is the wrong story
Matcha wants water at 70–80 °C, or 158–176 °F. Almost every guide agrees on that band, and almost every guide explains it wrong.
Boiling water does not scorch matcha. It over-extracts the bitter catechins, while the sweet, savory notes were already easy to pull at a lower temperature. The extraction literature points the same way: a matcha extraction study measured peak polyphenol and antioxidant yield at 80 °C, and a green tea brewing study found that catechin extraction tracks water temperature more closely than it tracks time.
The difference is not pedantry, it is trainable. “Fragile” tells a barista to be careful, which is not an instruction anyone can follow in the middle of a rush. “Over-extracted” tells them there is a dial and which way to turn it. Harsh cup, check the kettle before you blame the lot.
The contrast with hojicha makes the point sharper. Hojicha is brewed at 90–95 °C, or 194–203 °F, because roasting has already driven off most of what heat would over-extract. Same shelf, completely different band. Temperature belongs to the specific tea in front of you, not to a general rule about green powder.
Milk gets its own number, and it is the one most bars never write down: steamed milk belongs at 60–65 °C, or 140–149 °F. Two temperatures on the card, not one.
The two tools that make the number real
A dose is a fiction until something enforces it. Two pieces of hardware do that work, and both are cheap relative to the matcha they protect.
- A 0.1 g scale. Buy one, because a scoop swings 0.5–1.5 g on a 2 g target. That is not a rounding error, that is a different drink leaving the bar under the same menu name.
- A fine sifter, 60–100 mesh. A flour sifter is too coarse. A whisk cannot undo a clump once it hits water: the outer layer sets on contact and seals a dry core inside. Sifted matcha also re-clumps overnight under its own weight, so the sift happens at shift start, not the night before.
Which lot you are dosing is a separate decision from how you dose it, and it is covered in choosing a matcha for cafe lattes. This page is only about the numbers.
Pick one point in the range and hold it
The ranges above are real, not hedging. Sources genuinely differ, and 3 g and 4 g are both defensible for a 12 oz hot latte. What is not defensible is running both in the same week.
So pick one point inside the range, write it on the recipe card, and hold it. Consistency beats the exact number. The card needs six lines: grams by cup size, water temperature, paste volume in ml, milk temperature, whisk tool, whisk time. Laminate it and put it where the frother lives. To see what your grams cost per drink and in kilos a month, use the matcha drink cost calculator.
The card does not travel automatically when the lot changes. A Shizuoka Yabukita first flush and a Kagoshima Okumidori will not land in the same place at the same dose. Re-taste at your locked dose whenever a new lot arrives. That is the point of a fixed number: it makes a change in the tea visible instead of letting it hide inside barista variance. Once a lot holds your card, reorder that same tea as bulk matcha by the kilo, and the re-taste stays a check rather than a rewrite.
Where Hisa fits
A recipe card is an operations asset, and it only holds if the thing being dosed holds. That is the tradeoff most cafes end up eating. Go farm-direct for the quality and you inherit long lead times and lot-to-lot drift, so the card gets rewritten every few months. Buy for convenience and the supply is steady but nobody can tell you what is actually in the bag, so when the drink changes you have nothing to diagnose.
Hisa is built so that is not a choice. The line runs from the producer to a US warehouse, restocks ship domestically instead of clearing customs each time, and every lot arrives named by region, cultivar, and harvest. When a number stops working, you know which variable moved.
Tell us what you actually serve. Cup size, hot-led or iced-led, dairy or oat, and we will tell you which lot holds your card.