The short answer: A common starting point for a double espresso is 18 g of ground coffee, around 36 g in the cup (ratio 1:2) and a shot time of 25 to 30 seconds. This roughly matches the average shot in the Barista Guild of America (BGA) survey of 2017, and Sage gives the same time span in the Barista Express manual. It is not a standard: your manufacturer’s basket figure takes precedence, and you change only one variable per shot.
Your first shot, step by step
You need a precision scale, a clock and fresh beans. The order follows the extraction guide in the Sage Barista Express manual, which lists grinding, dosing, tamping and output time as the adjustable variables.
- Set the dose according to your basket. If nothing else is printed on it, 18 g is a workable start. In the BGA survey of 275 baristas, mostly from the USA, 66 per cent gave a dose of 18 to 20 g, most worked with 18 g baskets, and 93 per cent described their espresso as a double. For the large basket of the Barista Express, Sage gives 16 to 19 g, and for the Victoria, Lelit recommends 14 g for two cups. Where the starting value and the basket figure contradict each other, the basket figure applies, because the manufacturer designed its basket for that amount.
- Work out the target weight. At a ratio of 1:2, 18 g of ground coffee becomes an espresso of 36 g. The ratio is the average of the survey, the answers ranged from 1:1.5 to 1:2.5, and 82 per cent of respondents measure their yield by weight. Other quantities are simple to work out yourself.
- Grind only once the machine has heated up. There should be little time between the grinder and the shot; why is explained below under crema.
- Distribute the grounds and tamp them level. In the manual for the Classic E24, Gaggia writes that the grounds should be pressed evenly so that a level and smooth surface results, and that coffee residue on the basket rim should be removed. For the Barista Express, Sage gives a tamping pressure of 10 to 15 kg. Both statements aim at a bed that offers the water the same resistance everywhere.
- Time the shot from the press of the button. Sage expects the output to start after 8 to 12 seconds, to run slowly like warm honey, and the whole extraction to take 25 to 30 seconds. Because the start of flow falls within this span, Sage, as we read it, counts from the start; the manual does not say so explicitly. The BGA survey, in which 51 per cent gave 25 to 30 seconds, does not specify when timing starts at all. We count from the press of the button, so that the clock matches the Sage values.
- Change exactly one variable per shot. In their study of 2020, Cameron and colleagues list seven adjustable variables: grind size, coffee dose, water pressure, temperature, drink quantity, tamping force, and flow rate or time. If you turn two at once, you do not know afterwards which one had an effect.
- Write down the grind setting, dose, yield and time as soon as a shot works. After changing beans, check the time again: in 2003, Andueza and colleagues found different best grind sizes for two blends, fine for the normally roasted one and coarse for the torrefacto blend.
How to adjust the grinder itself and in what steps is explained in the guide Coffee grind size.
Why a finer grind runs more slowly
Espresso is made when water is forced under pressure through a compacted bed of ground coffee, taking soluble substances with it. The Deutscher Kaffeeverband describes it briefly in its brochure: the powder is lightly pressed in the portafilter, the water is forced through at high pressure, and the crema forms. For espresso, the association recommends finely ground coffee, for filter coffee a medium grind and for the French press a medium to coarser one.
The finer the grounds, the more goes into solution. In the same study of 2003, Andueza, de Peña and Cid compared three grind sizes on two blends with 80 per cent Robusta: solids, soluble substances and aroma compounds passed into the drink more strongly the smaller the particles were. For a pure Arabica, this is an extrapolation.
Fine grounds also slow the water. Cameron and colleagues name four ways to a slower flow: less water pressure, a finer grind, a more densely pressed bed or more coffee, also in combination. Smrke, Eiermann and Yeretzian described the mechanism behind this in 2024. A higher share of fine particles lowers the permeability of the bed, the flow falls, and the shot takes longer. Their recipe: 20 g of coffee, 40 g of drink, 9 bar, tamped with 20 kgF, on a Victoria Arduino Black Eagle.
So far it sounds as if you only need to grind fine enough. Cameron and colleagues found something different. Their model predicted, assuming uniform flow through the bed, that the extraction yield rises steadily with a finer grind. What they measured was a maximum: at a very coarse and at a fine setting the yield was lower, and at the fine setting the results were hard to repeat. The authors explain this with uneven flow, for example from too many fines, clumped particles or zones of different density in the bed. The grinding was done on an EK 43, brewed with 20 g to 40 g at 6 bar.
A range that Cameron and colleagues take from the SCA describes how much yield tastes good: coffee usually tastes best when 17 to 23 per cent of the dry mass is extracted, typically bitter above that and often sour below. The range comes from the control chart for filter coffee. It is carried over to espresso, and according to the same authors, espresso is the one that varies most among the common coffee formats.
How much pressure reaches the coffee bed
Many boxes say 15 bar. That is a genuine technical figure, but it describes the pump. For the E8, JURA gives a static maximum of 15 bar, that is, the pressure against a closed circuit; as soon as water flows, it drops. With the Classic UP, Gaggia separates the two itself: 15 bar nominal, 9 bar when brewing, held by a fixed-setting pressure relief valve. For the De’Longhi Dedica EC685, only the pump pressure is given. How much of it is present in the basket depends on the resistance of the coffee bed and, on many pump machines, on a pressure relief valve.
The sources see what should be applied to the grounds in a band around 9 bar. The historical SCAA definition, which SCA magazine 25 printed in February 2018, gives 9 to 10 atmospheres (one atmosphere is 1.013 bar). In the BGA survey the mean extraction pressure was 8.5 bar, the most common answer was 9 bar, and 52 per cent used pre-infusion. Illy and Navarini write in 2011 of 9 ± 2 bar. For the Victoria, Lelit gives 8 to 10 bar during extraction at 15 bar pump pressure.
One study compared pressure directly. In 2002, Andueza and colleagues brewed Arabica espresso at 7, 9 and 11 atm. According to the abstract, the samples at 9 atm showed good foam consistency (in the original, consistency of foam) and a high share of key aroma compounds, and the samples at 7 and 9 atm differed from those at 11 atm. So more pressure was not better in this study. Cameron and colleagues even worked at 6 bar, because their machine, set to 9 bar, clogged the bed at fine grind sizes.
A pressure gauge shows what is actually applied during the shot. On the Lelit Anna PL41TEM and the Lelit Elizabeth, the green zone runs from 8 to 12 bar according to Lelit. Our assessment: the bar figure on the box says little at the time of purchase about the pressure at the coffee bed. More revealing is whether the manufacturer states a brew pressure, builds in a pressure relief valve or supplies a pressure gauge.
Channelling: when the water takes its own path
Channelling is when the water flows preferentially through spots with lower resistance instead of passing through the bed evenly. In 2023, Lee, Smith and Arshad showed in a simplified model with two flow paths how this reinforces itself: an initially small difference in porosity and permeability grows through flow and extraction until one path is completely leached out. Below a critical fineness, extraction in the model falls with a finer grind, as in the experiment by Cameron and colleagues. This was not measured on a real puck; it is a calculation.
For flavour, something unpleasant follows. With uneven flow, some areas are extracted more than others, and the measured yield is only an average. Cameron and colleagues therefore expect a mixture of under- and over-extracted notes: the same espresso can be described as bitter and sour at once.
What helps against it is less well documented than the problem itself. Manufacturers ask for level tamping, see step 4 above. The force seems to matter less: Cameron and colleagues tried a range of tamping pressures and saw no notable change in shot time or yield; after that, a machine tamped with 98 N, around 10 kg. They noticed differences in flavour but did not quantify them.
WDT stands for Weiss Distribution Technique: you stir the grounds in the basket with thin needles before tamping. Normcore describes the purpose as breaking up clumps, loosening compacted grounds and distributing them more evenly, which should result in less channelling. That is a manufacturer statement. We have not found a peer-reviewed study on the effect of WDT on channelling. From the mechanics follows a trade-off that our German reviews of WDT tools describe: a needle breaks up clumps and at the same time pushes grounds aside, and the thicker it is, the more volume it displaces.
Two accessory aids belong here. In our German review of the Vicloon Tamper 51 mm, we advise adding a WDT tool for fine espresso grounds, because a tamper does not break up clumps deeper in the basket; that is our assessment, not a measurement. The scarlet espresso tamping mat is supposed to stop the portafilter from slipping during tamping, according to the manufacturer. The manufacturer itself does not claim that it prevents channelling.
The figures and who gives them
| Variable | Range in the sources | Who gives what |
|---|---|---|
| Dose, double espresso | 14 to 20 g | SCAA historical 14 to 18 g; BGA 2017: 66 % give 18 to 20 g; Sage Barista Express and Bambino Plus 16 to 19 g; Lelit Victoria 14 g; Cameron 2020 and Smrke 2024: 20 g in the experiment |
| Brew ratio | 1:1.5 to 1:2.5 | BGA 2017, mean 1:2; Cameron 2020 and Smrke 2024 each 20 g to 40 g |
| Shot time | 20 to 30 s, short shots 10 to 20 s | SCAA historical 20 to 30 s; BGA 2017: 51 % give 25 to 30 s; Sage 25 to 30 s, over 35 s too fine, under 20 s too coarse; Gaggia E24: under 20 s is a fault; Smrke 2024: turbo shots 10 to 20 s |
| Pressure at the coffee bed | 6 to 11 bar | SCAA historical 9 to 10 atm; BGA 2017 mean 8.5 bar; Illy and Navarini 2011: 9 ± 2 bar; Lelit Victoria 8 to 10 bar; Cameron 2020: 6 bar in the experiment |
| Brew temperature | 85 to 96 °C | SCAA historical 90.5 to 96.1 °C; BGA 2017 mean around 93.5 °C; Illy and Navarini 92 to 94 °C; Lelit Victoria 88 to 92 °C; Andueza 2003: 92 °C; Salamanca 2017: usual 85 to 95 °C |
| Tamping pressure | approx. 9 to 20 kg | BGA 2017: mostly 20 to 30 pounds (around 9 to 14 kg); Sage Barista Express 10 to 15 kg; Smrke 2024: 20 kgF; Cameron 2020: 98 N automatic, force without notable influence on time and yield |
| Extraction yield | 17 to 23 % | SCA according to Cameron 2020, from the control chart for filter coffee |
In four of the seven rows, a café source, a manufacturer statement and a study stand side by side, and they rarely match exactly. The SCAA values are explicitly a historical version; we have not found a current SCA standard for espresso. The BGA figures describe how baristas work in cafés, not how you have to work.
If the temperature can be set, we would start at 92 °C. In 2003, Andueza and colleagues chose this value for Arabica and for a Robusta blend on the basis of aroma compounds and sensory evaluation; according to the abstract, physico-chemical parameters, taste and mouthfeel were of little help in the choice.
The 25 to 30 seconds are not a law of nature. In 2024, Smrke and colleagues describe the growing spread of so-called turbo espressos of 10 to 20 seconds. In their own recipe, however, the highest yield came from shots over 40 seconds.
Faults: symptom, cause, what you change
The table follows mainly the troubleshooting section and the extraction guide in the Sage Barista Express manual. Grind size appears in almost every row, but never alone.
| What you see or taste | Cause according to the source | What you change |
|---|---|---|
| Output after 1 to 7 s, runs fast like water, thin pale crema, finished after 20 s at most; according to Sage bitter or sharp, thin and watery, classed as sour in the guide | Sage: ground too coarse, too little coffee in the basket or tamped too lightly, so too little extracted | grind finer; check dose and tamping |
| Output only after more than 12 s, drips, over 35 s; bitter, astringent | Sage: too fine, too much coffee, tamped too hard or basket clogged, so too much extracted | grind coarser; lower the dose; clean the basket |
| Gaggia Classic E24: shot under 20 s | Gaggia lists this as a fault and names grind size, tamping and coffee quantity | as in the first row |
| Pressure gauge above 14 bar, no coffee (Lelit Anna) | Lelit: too fine, too much coffee, pressed too hard or shower screen dirty | grind coarser, use less coffee, brush the shower screen |
| Bitter and sour at once, result hard to repeat | Cameron 2020: uneven flow, especially at a fine setting | distribute the grounds, tamp level, grind a touch coarser |
| No crema or hardly any | Sage: beans not fresh; Illy and Navarini: too long between grinding and the shot | freshly roasted beans, grind directly before the shot |
The rule of thumb behind this, sour means under-extracted and bitter means over-extracted, appears in the Sage guide and fits the SCA range above. Sage itself is not entirely consistent here: the extraction table one page earlier also names a bitter or sharp taste for shots that run too fast. For the second-to-last row it is unreliable, because both occur there at once. A temperature rule is missing from the table on purpose: we have not found in any primary source, for espresso, that sour is to be corrected with hotter water and bitter with cooler water.
If nothing runs at all despite correct values, the cause is more likely the machine; more on this is in our guide on portafilter problems. The last row of the table depends on the bean, which the next section covers.
Bean, roast level and crema
In its roast level overview, the Deutscher Kaffeeverband lists the dark roast level as ideal for espresso preparation, the medium roast for filter or café crème and the light one for filter. Illy and Navarini add two findings on crema in 2011: dark roasted beans contain more CO2 than medium ones, but the foam was most stable at a medium roast level (they cite Nunes and colleagues for this). We have not found a source that advises against dark roasts for espresso in general. Our assessment: roast level is a matter of taste, and the grind size is set according to the bean, not the other way round.
According to Illy and Navarini, the gas phase of the crema consists mainly of CO2, which forms during roasting and is trapped in the cell structure of the bean, and the foam volume depended strongly on the CO2 content of the powder. In a work by Barbera that they cite, over 70 per cent of the CO2 was released when grinding to 500 micrometres. That is why there should be no more than 0 to 30 minutes between grinding and the shot if the crema is to be preserved.
How to read roast date, variety and storage when buying is covered in our guide on bean basics. Beans from our comparisons are listed in our German coffee beans category.
Machine, grinder and tools from our reviews
If you do not have a machine yet, the basics are in our portafilter buying guide, with the models in our German portafilter category and our selection in our German best-of list for portafilter machines 2026. For this guide, what counts most about a machine is whether it leaves you the adjustable variables.
The Sage Barista Express (about €489.00 on Amazon.de according to our price box) has a conical burr grinder built in and a 54 mm portafilter. A razor blade for levelling comes in the box, and before full pressure, a pre-infusion at low pressure runs. Many figures in this text come from its manual. The Sage Bambino Plus (about €479.90) has no grinder; on the product page, Sage calculates with 18 g. Its double-wall baskets for pre-ground coffee regulate the pressure themselves. That makes it forgiving, but takes away the adjustment options this guide is about.
For the Gaggia Classic Evo Pro (about €499.52) with a 58 mm portafilter, Gaggia gives 15 bar nominal; the manufacturer has not published what value the pressure relief valve of the EU version is set to. The Classic UP, by contrast, holds a fixed 9 bar and shows it on an analogue pressure gauge, as described in the section on pressure.
Without a grinder in the machine, you need a grinder that grinds fine enough for espresso. The Eureka Mignon Specialita (about €431.90) has 55 mm flat burrs, stepless adjustment and a timer on the touchscreen. The Baratza Encore ESP (about €179.95) has a 40 mm conical burr grinder with 40 steps, of which 1 to 20 are finely stepped for espresso, a dosing cup for 54 mm with an adapter for 58 mm, and no timer. The German site covers grinders in more detail in the grinder guide and a best-of list for grinders 2026.
For the ratio in grams, you need a scale that reads to a tenth of a gram. The Maestri House mini coffee scale (about €29.99) resolves 0.1 g, weighs up to 2,000 g, has a timer with automatic start and stop detection and a rechargeable battery. Among the WDT tools, the Normcore WDT Tool V2.1 (about €25.99) has, according to Normcore, needles of 0.25 mm, 50 mm long, made of 304 stainless steel; the Amazon title says 0.23 mm. The Subminimal Flick (about €29.95) has retractable needles, 0.5 mm as delivered, with replacement sets of 0.35 mm available.
For the tamper, the diameter counts. For the Motta 8100/M (58 mm, flat stainless steel base, about €23.97), Motta sets the rule: inner diameter of the filter basket minus one millimetre. For 51 mm baskets there is the Vicloon for about €10.79. What else belongs on the work surface is covered in our guide on barista accessories and a best-of list for barista accessories 2026.
What we could not document
Some claims that circulate about espresso, and that in part appeared in older versions of this guide, we have not found in any reliable source. That 15 or 20 bar on the box describes the pressure at idle is not documented; what is documented is the static value against a closed outlet. For the claim that two machines with different bar figures deliver the same brew pressure or have the same pump, we found neither manufacturer documents nor studies.
Equally open: whether channelling can be reliably recognised at the spout (Cameron and colleagues infer uneven flow from the yield, not from the stream), where WDT historically comes from, and whether shot time is to be measured from the press of the button or from the first drop. The website of the Istituto Nazionale Espresso Italiano could not be retrieved by us, and of the two studies by Andueza and colleagues from 2002 and 2003, we read only the abstracts, not the full text.