The model: the espresso recipe turned into numbers

Every espresso is a recipe with three numbers: how much coffee goes into the portafilter, how much liquid lands in the cup and in how much time. This calculator turns them into the three indicators the coffee world uses to talk about extraction: the brew ratio — yield over dose, the celebrated "one to two" — the flow in grams per second, and the offset from the target you set for yourself. No sensory magic: weights, time and ratios — the common vocabulary that lets you repeat the good cup and correct the bad one.

The fourth number is the yield in grams, read straight from the scale: the model works in tenths of a gram precisely because the coffee scale shows them — and deserves them.

The four knobs: dose, yield, time and target

The dose runs from 14 to 22 grams: the range of the modern double basket, with 18 as the centre of gravity. The yield runs from 20 to 60 grams: weighed in the cup, not eyeballed — volume deceives, crema lies. The time runs from 15 to 45 seconds, from the first touch of the button. The ratio target runs from 1.5 to 3.0 in quarter steps: the recipe you want to hit — 2.0 the classic, 1.5 the declared ristretto, 2.5-3.0 the modern specialty lungo.

The yield scale: the same grounds, five drinks

An 18-gram dose, 27 seconds, 1:2 target; only how much you let into the cup changes:

The same dose produces five different drinks just by moving the point where you stop the shot: from the syrupy ristretto that stops at twenty grams to the border beyond three-to-one where espresso stops being espresso. The middle row is the model's factory recipe: thirty-six grams from eighteen, offset zero — the one-to-two that hits the target to the hundredth. And the last row also runs in the flow: at 2.22 grams per second the water is crossing the puck too fast.

The ratio: the number that does not look at the clock

The brew ratio has a property the calculator stages: it depends neither on the time nor on the target — it is the pure ratio of two weights. You can pull the same thirty-six grams in twenty seconds or in forty: the ratio stays 2.00, identical. It is the recipe's fundamental separation: the ratio decides the drink's style — how concentrated, how extracted — the time tells how you got there: the grind, the tamp, the health of the puck. Two different questions, two different numbers.

The flow: the witness of the grind

The second number is the most diagnostic: the average flow, yield over time. It has its own frozenness — it does not depend on the dose itself — and its alarm threshold: two grams per second. Below it, the water crosses the puck with the calm extraction requires; from there up, it is running — grind too coarse, dose too low for the basket, or a channel piercing the puck and stealing the flavour. The flow is the witness that does not lie: the ratio can be right by chance, the flow says whether the road was the good one.

The target and the offset: the compass of correction

The third number turns the calculator into a compass: the offset from the target, in hundredths of ratio. Negative: you are shorter than your goal — stop the shot later or grind coarser. Positive: you are longer — stop earlier or grind finer. Zero: the recipe is centred, and only consistency remains. The target is yours by choice: the model does not impose the one-to-two — it proposes it as the default because it is the counter's lingua franca, but the declared ristretto and the modern lungo are equally legitimate targets. What matters is having one: without a target, every cup is an accident.

The scale: the tool that changes everything

The whole model rests on a single gesture: weighing the cup. Volume deceives — crema puffs up, millilitres lie — and the tradition's "two fingers of coffee" produces ratios that dance by thirty percent from cup to cup. A coffee scale of a few euros under the cup, zeroed before the shot, turns espresso from ritual to recipe: weighed dose, weighed yield, timed shot — the three numbers of this page, repeatable identically tomorrow. It is the only purchase the model requires, and it is what separates chance from choice.

From the ristretto to beyond espresso: the four readings

The ratio classifies the drink. Below one and a half it is the ristretto: dense, sweet, with under-extraction around the corner — a choice to be made on purpose. Between one and a half and two point two it is the classic espresso: the window where the world works. Between two point two and three it is the modern lungo: the territory of light specialty roasts. Beyond three-to-one the drink leaves the territory: not forbidden, but the portafilter was not born for it — and it is more honest made with a filter.

The model's limits: what the three numbers do not say

The perimeter is declared: the count measures weights, time and ratios — not taste. It does not see the grind, which is the true knob behind the flow; it does not see temperature or pressure, which move the extraction at equal ratio; it does not see the roast, which decides which ratio performs best — dark compresses, light opens up. And it does not measure the true chemical extraction: for that you need the refractometer and TDS, the next step of the rabbit hole. These three numbers are the common base: the repeatable recipe on which everything else is tuned.

The dose scale: the ratio moves, the flow stays

Thirty-six grams in the cup in 27 seconds; only the dose in the portafilter changes:

The same thirty-six-gram cup changes nature according to the grounds that produced it: from the almost-lungo 2.57 of the reduced basket to the towards-ristretto 1.64 of the large one. But the flow column does not move: the grams per second do not know how much coffee was there — they only measure the liquid and the clock. It is the model's separation demonstrated in a table: the dose moves the style, not the flow — and whoever changes basket without recalculating the ratio is changing drink without knowing it.

The time scale: the ratio still, the flow that tells

Same 18-and-36-gram recipe; only the time it takes changes:

The ratio stays nailed at 2.00 on all five rows — the recipe is the same — but the flow crosses the whole diagnostic range: from the fifteen-second sprint, where the water pierces the puck, to the forty-five-second calm, where the fine grind chokes. The table's lesson is the page's most practical: the same ratio can hide five different extractions — and it is the time, not the ratio, that says which of the five you have in front of you.

The target scale: the same cup, five judgements

The third scale does not touch the machine: only the declared goal changes. The same 2.00 extraction is worth plus fifty against the 1.5 ristretto target, exactly zero against the one-to-two, minus one hundred against the 3.0 lungo. It is the reminder that the offset is a relative measure: it does not say whether the cup is good — it says whether it is the one you wanted. Changing target without changing recipe transforms the judgement, not the coffee: first you choose the style, then you judge the offset — never the other way round.

The grid's extremes: from 0.91 to 4.29

The model's minimum is almost a declared absurdity: twenty-two grams of grounds for twenty of beverage — ratio 0.91, less liquid than coffee, a concentrate beyond the ristretto that almost no machine pours willingly. The maximum is the opposite: sixty grams from fourteen in fifteen seconds — ratio 4.29 with a flow of 4 grams per second, a torrent no healthy puck produces. Between the two extremes runs the portafilter's whole territory — and the fact that both exist in the grid is the point: the calculator also measures the wrong extractions, because that is where it is needed most.

The dial-in protocol: one knob at a time

The model suggests the method serious counters use to tune a new coffee. Fix the dose — the basket's — and never touch it again. Choose the target according to the roast: 2.0 for the medium, longer for the light. Pull and weigh: the offset tells you where you are. Correct with the grind, not with the dose: finer if the flow runs, coarser if it chokes — one notch at a time, one variable at a time. In three or four cups the offset converges to zero, and from there the work is only defending consistency: same dose, same grind, same gesture.

The strategies of the repeatable cup

Four concrete moves follow from the model. The scale under the cup, always: without weight there is no ratio, and without ratio every correction is a shot in the dark. One convention for the time: from the button, pre-infusion included — and never change it, because comparisons hold only at a constant yardstick. One variable at a time: whoever changes dose and grind together will never know which of the two worked. And the notes: date, coffee, dose, yield, time, offset — six numbers per row, and it is the notebook that turns attempts into a house recipe.

Frequently asked questions

What is the brew ratio?

The yield in the cup divided by the dose in the portafilter, both in grams: 36 grams from 18 make 2.00 — the famous one to two. It is the number that defines the drink's style, from the ristretto below 1.5 to the modern lungo towards 3.

Why weigh the cup instead of watching the volume?

Because crema puffs up and millilitres lie: the same weight can occupy volumes thirty percent apart. The scale under the cup, zeroed before the shot, is the only way to make the recipe repeatable.

What is the right ratio to start with?

The one-to-two: dose times two in the cup, in about 25-30 seconds. It is the counter's lingua franca and the calculator's default — from there you explore towards the ristretto (shorter) or the modern lungo (longer) according to roast and taste.

What does the flow in grams per second tell me?

How the water crossed the puck: below 1-1.5 g/s the extraction is calm, towards 2 it is running. It is the diagnostic number: a right ratio with high flow tells of a grind too coarse or channelling — right by chance, not by choice.

Why does the ratio not depend on the time?

Because it is the ratio of two weights: the same 36 grams from 18 make 2.00 in twenty seconds as in forty. The time tells the journey (grind, tamp), the ratio tells the destination: two different questions.

How do I use the offset from the target?

As a compass: negative, you are shorter than the goal — stop later or grind coarser; positive, you are longer — stop earlier or grind finer; zero, the recipe is centred. Fifty hundredths of offset is a serious correction, ten is fine-tuning.

When does the running extraction trigger?

When the flow reaches 2 grams per second: water too fast through the puck. The classic causes: grind too coarse, dose too low for the basket, uneven tamp or channelling. The correction almost always starts from the grinder.

Is time counted from the button or from the first drop?

The model uses the first touch of the button, pre-infusion included: it is the most repeatable method because it does not depend on the eye. What matters is choosing one convention and never changing it — the gap between the two is a constant 5-8 seconds.

Which dose do I choose in the portafilter?

The basket's: modern doubles are born for 18 grams, with 14-22 variants. The practical rule: the basket decides the dose, taste decides the ratio — changing the dose without changing the basket alters the extraction bed.

Is the ristretto a failed espresso?

Not if it is chosen: a ratio below 1.5 with a grind designed for it gives a dense, sweet cup. It failed only when it is suffered — the too-fine grind that chokes the shot and stops the cup short with the aggressive acidity of under-extraction inside.

Is the modern lungo like the bar lungo?

No: the counter lungo is often an espresso stretched past its time, bitter. The modern lungo is a recipe: ratio 2.5-3 with a light roast and a finer grind, designed to open acidity and aromas. Same name, two different drinks.

Do I need a refractometer to use this calculator?

No: refractometer and TDS measure the chemical extraction, the next step. These three numbers — ratio, flow, offset — are the base everyone needs: scale and timer suffice, and they are ninety percent of repeatability.

Updated July 2026