How to dial in espresso, when the recipe everyone gives you is a survey result

Search for how to dial in espresso and six different pages will tell you the same thing. Eighteen grams in. Thirty-six grams out. Twenty-five to thirty seconds. Sour means grind finer, bitter means grind coarser, change one thing at a time.

I read the top results for four of these searches and none of them cites a source for any of it. One credits “professional coffee institutes” without naming one. One lists its references and among them is the consensus of a subreddit. One is written by a machine manufacturer, and its citations are two blogs.

So I went looking for where the numbers came from, and the answer turns out to be worth the trip. The recipe is not a standard. It is the average of a survey — 275 baristas, asked in March 2017 what they actually do, published by the Specialty Coffee Association the following February under the title Defining the Ever-Changing Espresso.

And the definition that survey was measured against contains one clause that quietly explains why dialling in feels circular. It defines the correct grind as whichever grind produces the target time.

The recipe everyone gives you, and where it came from

In that 2018 article the SCA reproduces its own older definition, and calls it the heritage Specialty Coffee Association of America definition:

“Espresso is a 25–35ml (.85–1.2 ounce [×2 for double]) beverage prepared from 7–9 grams (14–18 grams for a double) of coffee through which clean water of 195°–205°F (90.5°–96.1°C) has been forced at 9–10 atmospheres of pressure, and where the grind of the coffee is such that the brew time is 20–30 seconds. While brewing, the flow of espresso will appear to have the viscosity of warm honey and the resulting beverage will exhibit a thick, dark golden crema. Espresso should be prepared specifically for and immediately served to its intended consumer.”

Read the bolded clause twice, because everything downstream inherits it.

The definition does not specify a grind size. It cannot — grind size in espresso has no unit anyone agrees on, no reference scale, no way to write it down that survives moving between two grinders. So the definition does the only thing available to it and defines grind as a function of time: the right grind is the one that gives you 20 to 30 seconds.

Then every guide written since turns that around and tells you to adjust the grind until you hit 20 to 30 seconds.

That is not a method. It is the same sentence read in two directions. The standard says correct grind produces the target time; the guide says target the time to find the correct grind. Nothing in the loop tells you whether the coffee is any good, which is why every guide has to bolt a taste step onto the side of it.

It is worth saying plainly that this is not a live standard. The SCA called it heritage in 2018, and the piece that quotes it exists precisely because the association was collecting data to replace it. But the numbers escaped the document, and they are still in circulation eight years later with none of that context attached.

The rest of the definition has aged in its own way. It describes the flow as having “the viscosity of warm honey” and the result as carrying “a thick, dark golden crema” — a visual specification, in a document with no instrument in it, for a foam that turns out to be a gas gauge rather than a quality gauge. Two of the definition’s three sentences are about what the shot looks like.

What 275 baristas actually do

The survey was built by the Barista Guild of America with SCA research staff: 26 questions, circulated in March 2017, 275 respondents, 76% of them in the United States, with the largest international contingents in the UK, Singapore and Canada. Fasman states the purpose without hedging — the results “would be critical information in crafting new best practices and standards.”

Here is what came back, next to the definition it was measured against.

One value in that definition still describes practice, and it is the water temperature. Everything else has moved, in some cases by a factor of two.

The survey has a few more findings that no dial-in guide I read mentions:

• 93% call what they pull a double. At 18 to 20 grams it is not the heritage double, which topped out at 18. The word survived; the quantity behind it did not.
• 82% measure yield by weight rather than volume, which is the single biggest methodological change between the definition and the practice.
• 52% use pre-infusion, a variable the definition does not contain at all.
• Most use an 18 gram basket. Spouted and bottomless portafilters were almost exactly tied at about 38% each.
• On tamping, most reported 20 to 30 pounds — but 22% answered “other”, and their comments were that they had never measured it, that a mechanical tamper did it, or that they tamped “until the puck was evenly dense.”

That last one deserves a moment. Nearly a quarter of professional baristas could not put a number on the variable that dial-in guides treat as a discipline problem.

So when a page tells you 18 in and 36 out as though it were a rule: it is a rule the way “the average American household has 2.5 people” is a rule. It is a description of a population. The article that published it is called Defining the Ever-Changing Espresso, and it ends by promising more surveys, because the authors expected the numbers to keep moving.

Three authorities, three different drinks

The heritage definition is not the only document with numbers in it. There is a live third-party certification for espresso, and there is a body of peer-reviewed experiment. Putting them side by side is more useful than any of them alone.

The Italian figures come from the Istituto Nazionale Espresso Italiano’s own booklet. The institute was founded in 1998 after three years of research with the International Institute of Coffee Tasters and the Centro Studi Assaggiatori, and its product conformity certificate — CSQA number 214, dated 24 September 1999 — is issued under ISO 45011 by a third-party body. This is a real certification with a paper trail, and it describes a drink less than half the size of the one the survey respondents pull.

Two things stand out in that table.

Nobody specifies the grind. The live certification requires a “qualified grinder-dispenser” and gives no grind figure whatsoever. The retired definition defines it by the clock. The survey did not ask. The only source that treats grind as a first-class variable is the peer-reviewed model — which is also the only one that tells you to stop watching the clock.

The experimental work lines up with the Italian column, not the modern one. The temperature that won Andueza’s trials was 92 °C, inside the Italian band and at the bottom of the heritage one. The pressure that won was 9 atmospheres, which is the Italian figure. The ratios Andueza compared were 6.5 and 7.5 grams per 40 ml of water, which is the Italian shot’s territory and nowhere near 1:2. The measurements that exist were made on a different drink from the one most people are trying to dial in.

The 25-second rule has a tolerance nobody quotes

While reading the Italian booklet I found a small error that has propagated widely, and it is the kind that is easy to check and worth checking.

The institute’s parameter table reads, verbatim:

Percolation time is 25 seconds plus or minus five. It is widely repeated as plus or minus two and a half — including by the summary a search engine generated for me while I was looking for the document. The ±2.5 in that table belongs to the volume, one row down. Two adjacent rows, one tolerance borrowed from the other, and it has been copied ever since.

The correction matters more than it looks. At ±5 seconds, the certification’s window is 20 to 30 seconds — the entire band the heritage definition allows. A shot at 21 seconds and a shot at 29 seconds are both certifiable Italian espresso. The clock, in the one place it is genuinely enforced, is not fine-grained at all.

Finer is not a direction. It is half of a curve.

Every guide states the same rule as though it were physics: grind finer to extract more, coarser to extract less.

It is not physics. It is one side of a curve.

Cameron and colleagues published the first mathematical model of espresso extraction in Matter in 2020, tested against experiment on a San Remo Opera with a Mahlkönig EK43, 20 g ridgeless baskets and a recipe of 20.0 ± 0.5 g in, 40.0 ± 0.5 g out at 92 °C, with shots discarded if the beverage mass fell outside a ±1 g tolerance. What they measured was this:

“measurements show a peak in the extraction yield versus grind setting relationship, with lower extraction yields at both very coarse and fine settings. This result strongly suggests that inhomogeneous flow is operative at fine grind settings, resulting in poor reproducibility and wasted raw material.”

And:

“We identify a critical minimum grind size that allows for homogeneous extraction. Below this setting, a counterintuitive reduction in EY and increase in variability is observed.”

Extraction yield rises as you grind finer, reaches a maximum, and then falls. Past that point the bed clogs unevenly, water finds channels through it, and the average extraction drops while the shot-to-shot spread grows. Grinding finer to fix a sour shot works on the coarse side of the peak and makes things worse on the fine side — and nothing in the cup tells you which side you are on, because both sides taste under-extracted.

The consequence is sharper than that, and the paper says it outright:

“one can obtain two 22% EY shots by keeping the brew ratio fixed and setting the GS to either 1.3 or 2.0. The chemical composition of the faster shot cannot be the same as the slower shot… This result does not undermine our use of EY, but rather illustrates that the barista indeed needs to taste the coffee, rather than measure its solvated mass.”

Two grind settings, the same extraction yield, different chemistry, different flavour. A number that can be reached from two directions is not a target you can navigate to.

There is a small irony sitting underneath that. Cameron’s paper reports the target band as the SCA advising that coffee “most frequently tastes best when the proportion of extracted dry mass is in the range 17%–23%” — but the figure everyone else repeats is 18 to 22%, and when I read SCA Standard 310-2021 for an earlier piece on that standard, neither range appeared in it. Three sources, three versions of the same window, and the document they all point at contains none of them. I am quoting Cameron’s attribution here as Cameron’s, not as the SCA’s, because I could not find it in an SCA document to check it against.

Your grinder is noisier than your technique

There is a second problem with change-one-thing-at-a-time, and it is that the thing you change is not the only thing that moves.

Severini and colleagues studied the variance of espresso extraction variables in a 2016 paper whose subject is literally espresso as served in coffee shops. Their finding:

“The major source of information was contained in the grinding grade, which accounted for 87–96% of the variance of the experimental data. The variability of the grinding produced changes in caffeine content in the range of 80.03 mg and 130.36 mg when using a constant grinding grade of 6.5.”

Read the second sentence carefully. At a constant grinder setting — nothing touched, nothing adjusted — caffeine content ranged from 80 to 130 milligrams. That is a swing of more than 60% with no change to the variable you think you are controlling.

Cameron’s team put the same point in terms of blame:

“Yet, the origin of this inconsistency has traditionally, and incorrectly, been attributed to human variations. This study’s mathematical model, paired with experiment, has elucidated that the grinder and water pressure play pivotal roles in achieving beverage reproducibility.”

This is the finding that most directly contradicts the advice you will read everywhere. When two shots differ, the standard guidance is to look at your puck prep — your distribution, your tamp, your technique. The modelling says the dominant terms are the grinder and the pressure, and calls the human-error explanation incorrect.

Which also means the one-variable-at-a-time protocol has a hidden assumption: that the grinder is deterministic, so any difference between shot one and shot two must be the variable you changed. It is not, and it is not. Some of what you are reading as a result is noise from the machine, and if you adjust in response to it you will chase it around the dial.

What the model says to do instead

Cameron’s paper is not a critique. It proposes two concrete protocols, and the second one is the reason this piece has the title it has.

Affiliate links. Some product names below link to Amazon. If you buy through one I earn a commission at no extra cost to you, and it never changes which product gets named or what the measurements say. Full disclosure.

Strategy one — find the peak, then cut the water. Locate a shot you actually like. Then hold everything else and move the grind alone until extraction is at its maximum, judged either with a refractometer or, in the authors’ own words, “by simply tasting the coffee.” Then reduce the water to bring the extraction back to where your good shot was. The result is a smaller, more concentrated, more repeatable cup.

Strategy two — dose down and grind coarse.

“a barista is able to achieve highly reproducible espresso with the same EY as the 20 g espresso by reducing the coffee mass to 15 g and counter-intuitively grinding much coarser… This modification may result in very fast shots (<15 s), a reduction in espresso concentration, and a different flavor profile.”

Twenty-five percent less coffee, the same extraction, better reproducibility, and shots under fifteen seconds. They ran it in an actual café: coffee valued at $0.53 per 20 g, a saving of $0.13 per drink, $3,620 a year, and “the shot times were routinely reduced to 14 s.”

Then comes the sentence that connects all of this back to the recipe at the top of the page:

“The Specialty Coffee Association espresso parameters mandate that the extraction should take 20–30 s; we speculate that this might be partially responsible for the prevailing empirical truth that most coffee is brewed using grind settings that cause partially clogged/inhomogeneous flow.”

The time target may be the thing pushing people into the clogged regime. If that is right — and the authors flag it as speculation, so I will too — then the rule everyone follows to make good espresso is also the rule wasting their coffee. The paper’s own summary is blunt about the goal: “Optimization of espresso parameters to minimize coffee waste.”

The part of this that nobody quotes

If I stopped there I would be doing the thing this piece is complaining about. The paper has a caveat, and it is a real one:

“It is clear that espresso made at 22% EY in the partially clogged regime tastes more ‘complex’ than a fast 22% EY obtained using the optimization routine presented in Figure 6.”

The optimised shot is cheaper and more consistent and may taste less interesting. The authors’ proposed fix — blend a low-extraction shot with a high-extraction one to recover the complexity — is something they themselves expect “only the most enthusiastic practitioners would consider.”

So this is a trade, not an upgrade. If what you want from espresso is the same drink every morning with less coffee, the down-dose route is well supported. If what you want is the most interesting thing that cup can be, the messy clogged shot may still win, and the people chasing it are not wrong.

Nobody has published a 1:2

The 1:2 ratio is treated as bedrock. I went looking for the study behind it.

There is exactly one peer-reviewed paper on espresso coffee-to-water ratio, from the same Navarra group as the pressure and temperature work. Here is what they compared:

“only some sensory attributes, such as bitterness, astringency and burnt, acrid and earthy/musty flavours were proposed as relevant to the selection of 6.5 g 40 mL⁻¹ or 7.5 g 40 mL⁻¹ in conventional roasted coffees (Arabica 100% and Robusta blend), and 6.5 g 40 mL⁻¹ in torrefacto roasted coffees.”

Six and a half or seven and a half grams per forty millilitres. Those are ratios of about 1:6 and 1:5 — and even allowing for the fact that they are counted on water going in rather than beverage coming out, a distinction that changes what a ratio means, they are not in the same neighbourhood as 1:2. The published experiment on espresso ratio was run on the Italian shot.

The same abstract carries a second finding that gets no airtime at all:

“The optimisation of other technical parameters in previous studies seemed to minimise the influence of an increase in the coffee/water ratio on the extraction of soluble and solid compounds.”

With temperature and pressure set properly, raising the ratio did comparatively little to how much came out of the coffee. What separated the two ratios was not extraction but a short list of sensory attributes — bitterness, astringency, burnt, acrid, earthy. Ratio, in the one place it has been tested, is a flavour lever rather than the master control the guides make it.

None of which makes 1:2 wrong. It is a widely used, perfectly sensible convention, and 275 baristas converged on it. It is just not a finding, and no page that hands it to you as one can point at where it was found.

Sour and bitter are not a compass

The taste rule is the last piece of the standard advice, and it is the one I can support least.

The rule says sour means under-extracted and bitter means over-extracted, so grind finer for the first and coarser for the second. The nearest thing to a controlled test of that mapping is Batali, Ristenpart and Guinard’s 2020 study, which placed nine coffees across the brewing control chart — strengths of 1.00, 1.25 and 1.50% TDS crossed with extractions from 16 to 24% — and put them in front of twelve trained panelists. Their correlations:

“Five attributes increased with TDS only — Bitter, Astringent, Rubber, Earthy, and Viscous… attributes like Sour, Citrus, Dark Green, Smoky, Brown Spice, and Fermented all increased strongly with TDS but had a negative correlation with PE.”

Bitterness tracked strength, not extraction. Sourness went both ways — down with extraction, which is the folk rule, but also strongly up with strength, which the folk rule has no room for.

I have to be careful here, and more careful than the finding is exciting. That is a drip study. It was run between 1.00 and 1.50% TDS. Espresso lives around 8 to 12% — an order of magnitude outside anything they tested, and I have no basis for carrying their numbers across that gap. What I can say is narrower and still worth saying: in the one controlled sensory experiment that put strength and extraction on separate axes, both of the tastes that dial-in guides read as extraction signals moved with concentration too. A shot can be bitter because it is strong. In a 1:2 espresso, it is very strong.

Temperature muddies it further. Mark Prince’s advice at CoffeeGeek is that a sour shot means the machine is running too cold and a bitter one means too hot — and Batali’s team, summarising Andueza’s espresso work, report the opposite direction:

“An investigation of espresso coffee systematically varied the water temperature between 88 and 98 °C and found that hotter water predictably increased TDS and PE, and yielded more acrid, roasty, bitter, and sour attributes”

That sentence is Batali’s characterisation of Andueza, not Andueza’s own words — it is in the results of a paper I could only read the abstract of, and I am not going to pretend otherwise. What Andueza’s abstract does say is that they tested 88, 92, 96 and 98 °C on three coffees and selected 92 °C for all three, including the one whose overall acceptability pointed at 88 until the aroma compounds and a “not hot enough” perception overruled it. And it says something that ought to give every taste-led guide pause:

“physicochemical, taste and mouthfeel parameters were not very useful for selecting the water temperature”

The variable they could not choose by tasting is the one CoffeeGeek recommends choosing by tasting.

If you want a published description of what a balanced shot is supposed to taste like, the Italian institute has one, and it is refreshingly free of adjectives:

“Its taste is round, substantial and smooth. Sour and bitter are well balanced and neither one prevails over the other. Astringency is absent or barely perceptible.”

That is a certifying body’s definition of balance: not the absence of sour, not the absence of bitter, but neither one winning. Which is a better target than a direction on a dial.

Three shots, and the man who coined it

The other promise in every dial-in guide is speed. Three shots. Twenty minutes. A protocol you can run before work.

That target has a single origin. It is Mark Prince’s, at CoffeeGeek, and his credentials are real — a judge certified for the Canadian, US and World Barista Championships in both sensory and technical fields, who founded the site in 2001. He describes how the mantra formed: attempt to dial in any coffee, on any espresso grinder, in three shot pulls; take 60 grams of coffee and have the grinder tuned by the time it is gone.

Here is how he describes its success rate.

“Did it happen every time? Hell no. For many years, it was a goal, not a fait accompli.”

“I’ve been doing this for over 20 years now, and I still don’t get it at times. Indeed, the inspiration for this article was a rather excellent decaf espresso received recently that took me six dial in shot pull attempts.”

“it’s not about getting a perfect shot in three shot pulls, and everything about being a better barista.”

A World Barista Championship judge, twenty-five years in, needed six shots on a recent bag — and wrote the article to say so. The pages that repeat his number drop the paragraph where he explains it is aspirational.

His piece also disagrees with the modern consensus in a way worth keeping. On grinders with coarse steps he recommends moving the dose rather than the grind, up to 1.5 g either way, because “on some grinders, the difference between one ‘click’ on a grind adjustment can be the difference between a double shot that runs for 15 seconds to produce 60ml, and 40 seconds to produce that same 60ml.” Every guide that tells you to lock the dose and touch nothing but the grind is assuming a grinder fine enough to be worth touching. If yours has four usable espresso settings, that advice is not written for your machine.

How precise does the dose actually need to be

Since the protocol below asks you to fix a dose, it is worth knowing how fixed it needs to be. The published tolerances cluster tightly:

Every one of them is in the low single digits to low tens of percent. None is in tenths of a gram. A 0.1 g kitchen scale resolves five times finer than the tightest of them, which is the whole argument about scales in one line: resolution stopped being the problem a long time ago.

And note what Cameron’s team did with the number. They did not require the shot to hit 40.0 g — they recorded what it actually was and used it: “Exact beverage masses were included in each calculation of EY.” The scale’s job is bookkeeping, not gatekeeping.

If you want the dose your basket is rated for, that comes from the basket rather than from any standard — there is no published gram figure for a double shot, only VST’s rated windows and the Italian 7–9 g single that, doubled, produces the 14–18 g the internet repeats.

Getting your own number

Here is what I would actually do, built from the sources above rather than from the consensus. It is not faster than the guides promise. It is honest about which parts are supported.

Start from taste, not from time. Cameron’s routine begins by locating “a so-called tasty point (i.e., an espresso shot that tastes good to the barista).” Everything after that is anchored to a shot you liked, which means your first job is to find one, not to hit a number.

Fix the dose to your basket’s rating and leave it. Unless your grinder has too few steps to be useful, in which case Prince’s ±1.5 g dose adjustment is the documented workaround.

Move the grind alone, and taste every shot including the bad ones. You are looking for the point where extraction stops improving. Grind finer until the shot stops getting better and starts getting thin and harsh at once — that is the far side of the peak, and it is the signal that you have gone past the critical minimum grind size, not that you need to go further.

Record the time; do not steer by it. Write it down, because it tells you what the grind did. But the model’s recommendation is explicit — navigate “using only mass of coffee and mass of water as independent variables.” The clock is an instrument, not a destination. And if it lands at 22 seconds or 29, the one live certification for espresso counts both as in tolerance.

Then choose which trade you want. Bring the water down to concentrate the shot, or bring the dose down and the grind coarser to make it cheaper and more repeatable. Both are in the paper. The second one saves 25% of your coffee and may cost you some complexity.

Expect more than three shots. The person who invented the three-shot goal misses it, and says so.

What none of this gives you is a number to type into a grinder, and that is the honest end of the story. Espresso is the one brewing method where the dominant variable has no unit, no reference scale and no way to be written down — which is why the definition everyone inherited had to describe it as “the grind of the coffee is such that the brew time is 20–30 seconds,” and why eight years of guides have been quoting the answer to a question the document never actually answered.