Puck Prep & Channeling: Extracting Evenly
Sour and bitter in the same cup?
You did everything the same as yesterday — and yet today's espresso tastes sharp and dry-bitter at the same time. Almost always, channeling is behind it: the water finds a channel through the coffee puck instead of flowing through evenly. Here you can see how distribution, tamping, grind size and a puck screen change that.
The controls: Three relative scales from 0 to 100 without a unit — distribution (WDT), tamping (how level, not how hard) and grind size (direction coarse–fine) — plus the puck screen switch.
The results: Evenness and channeling risk are indicators from 0 to 100 from our model, not measurements. They show the direction, not how your espresso will taste.
The basis: The ranking of the levers (distribution first, then tamping, then grind size) is well supported; the numbers behind it are estimated. No guarantee — details right below the tool.
In more detail: controls, model and sources
The controls in detail: All three are relative scales from 0 to 100 without a unit — no grams, no kilograms, no micrometres. Distribution (WDT / grooming): 0 means clumpy and uneven, 100 means finely loosened and level. Tamping: this is the most common mix-up at home — the number is not a force. “Tamping 70” doesn't mean 70 kilos or 70 newtons, but: already pressed fairly straight and evenly, though not yet perfectly level. 0 would be a crooked, uneven tamp, 100 a completely level one. Grind size: 0 is coarse, 100 is very fine. Again, these aren't micrometres, just the direction fine–coarse. Puck screen is a switch, on or off.
The results in detail: Channeling risk and evenness are indicators from 0 to 100, not measurements and not lab results. They show you the direction your puck is moving in, not how the espresso will taste. The two numbers are the same information seen from two sides: evenness is simply 100 minus channeling risk. In the picture you also see whether the tool draws zero, one or two channels — that's an illustration of the indicator, not a prediction of where a channel would form in your basket. The two small bars “tends sour” and “tends bitter” are model values too: the more channeling, the more of both at once; grinding coarse pushes additionally towards sour, very fine towards bitter.
The basis in detail: The risk is made up of four contributions: poor distribution counts with a weight of 0.62, a crooked tamp with 0.50, a very fine grind with up to 0.25, and a puck screen on top subtracts a flat 0.12. Grind size only starts counting at all from level 55. The ranking of these levers is well supported — distribution first, then tamping, then grind size; this matches the espresso modelling by Cameron and Hendon (2020). The numbers themselves are not. 0.62, 0.50, 0.25 and 0.12 are estimated weights, not measurements, and the threshold of 55 for grind size is estimated as well. They put the levers in a plausible order; they don't quantify them. A model that shows the direction — and it shows that reliably.
What the picture shows: A cross-section through basket and puck, schematic and heavily simplified. Blue dots are water, arrows show direction and speed. Where a channel forms, the water runs faster (outlined in red) and dissolves too much there (dark, bitter); the rest gets too little water (light, sour). The position of the channel in the picture depends on the control setting, not on your basket.
Correction note (16.10.2026): The earlier version of this page called the measured effect of a puck screen “small and inconsistent”. Two measurement series (Ribes 2021 on a Decent DE1, Clive Coffee 2025 on a La Marzocco Linea Mini) show around 2 to 3 percentage points more yield, mainly because the puck edge extracts more evenly. We've made the text below more precise; the model weight of −0.12 remained unchanged.
The Puck Prep Simulator is loading … If nothing appears, please reload the page — or have a look at barista tools.
This is a simulator. It works with a simplified model and shows you relationships — not measurements and not an assurance of any kind. What ends up in your cup depends on your machine, your water, your beans and the way you work.
Our recommendations are based on experience from consultations, courses and the showroom. They are non-binding and provided without guarantee. We accept no liability for damage arising from their use. When in doubt, what your manufacturer writes in the instruction manual applies.
EstimateStraight to the results: Channeling risk and evenness are indicators from 0 to 100 from our model — not measurements. They come from weights we have estimated, and the channels in the picture are drawn, not measured.
What happens during channeling
Water always takes the path of least resistance. If the puck is uneven — clumps of fines, loose spots, a crooked tamp — the water flows preferentially through these weak points. In the channel, extraction is locally far too high (bitter, astringent); the rest of the puck stays under-extracted (sour, hollow). The result: both at the same time in one cup.
The mathematical modelling of espresso shows why: very fine particles tend to clump and create density differences that favour uneven flow. Even distribution and a level tamp are therefore the most effective levers for a clean extraction — more effective than most people think. In our roastery we deliberately grind very fine and extract slowly; that's exactly why preparation isn't a side step for us but the start of every shot.
One point needs to be brought together, because this page could answer it differently in two places: the puck screen. The tool above puts it at 12 of 100 points less channeling risk. In the calculation, the screen thereby replaces roughly what you'd also achieve with 19 points better distribution or 24 points better tamping. It's the weakest of the four levers — and the 12 points are estimated, not measured; the 19 and the 24 are calculated from that. What was measured elsewhere (see below): more yield, mainly at the puck edge. So the screen doesn't replace clean distribution; it complements it.
Peer-reviewedSource: M. I. Cameron, C. H. Hendon et al. (2020), “Systematically Improving Espresso: Insights from Mathematical Modeling and Experiment”, Matter 2(3), 631–648 (Cell Press), doi:10.1016/j.matt.2019.12.019. Additionally: review article “Culinary fluid mechanics and other currents in food science” (2022). The weights in the simulator's calculation (0.62 for distribution, 0.50 for tamping, 0.25 for grind size, −0.12 for the puck screen) are not derived from these — they are reasoned assumptions.
How hard do you need to tamp?
Less important than many think. Level counts, not hard. Once the puck is compacted, more force hardly changes anything: in 2025, the Swiss roastery Kaffeemacher pulled 150 espressos with a food technologist at 10, 15 and 20 kg of tamping pressure (automatic tamper, refractometer) — no detectable difference in flow time and dissolved solids (TDS). Below 10 kg was not measured; the authors recommend at least around 6 kg. A measurement series on compaction shows the same picture from the other side: about half of the compaction happens at very low force already, after which the gain quickly gets smaller. That's why pressure-regulating and self-levelling tampers don't rely on force but on the base sitting straight — the same, shot after shot.
Community measurementKaffeemacher, “Welchen Einfluss hat der Anpressdruck beim Tampen?” (19.08.2025), kaffeemacher.ch, accessed 16.10.2026. R. McKeon Aloe, “The Diminishing Returns of Tamping for Espresso”, Towards Data Science, link, accessed 16.10.2026. Neither re-measured by us.
Clean puck prep — step by step
- Dose & level: Consistent dose by weight, distribute the coffee evenly in the basket — no mounds, no holes.
- Loosen & distribute (WDT): Break up clumps with fine needles and distribute evenly. This is the biggest lever against channeling. John Weiss described the technique in 2005 on the Home-Barista forum — back then with a single dissecting needle; today it's usually several needles of 0.1 to 0.5 mm.
- Tamp level: Straight and even. How hard hardly matters once the puck is compacted (see above). A crooked tamp is a channel waiting to happen.
- Optional puck screen: A fine mesh on top of the puck distributes the incoming water more evenly at the top and keeps the shower screen clean.
- Look & taste: With a bottomless portafilter you see spurts and channels straight away. Sour and bitter at the same time? Then back to step 1 — no right or wrong, just more even. And always: dose in and yield out in grams, not by the clock. If you like it, you're right.
CommunityOn the origin of WDT: Daily Coffee News, “What is WDT in espresso? We talked to its creator John Weiss” (14.12.2022), link, accessed 16.10.2026.
Our experience from the roastery and our cafés: we grind fine, start at 14.5 grams and work our way up with the dose, usually to 17 grams, at a ratio below 1:2. More coffee at the same ratio can soften acidity and bitterness; extraction and taste do not behave linearly.
What helps against channeling
The most effective tools are a WDT tool for even distribution, a well-fitting tamper (or an automatic tamper like Puqpress for a consistently level puck) and a puck screen for even water distribution from above. Which tool fits which step is something you put together in the Puck Prep Workshop; the physics behind WDT and self-levelling tampers is explored in this article. How over- and under-extraction then feel in the cup is explained in our sensory basics; you can play through the numbers in the Extraction Simulator. Questions about this are answered by Jackson, our AI advisor, around the clock.
Frequently asked questions
What is channeling?
Channeling means the brew water finds channels through the coffee puck instead of flowing through it evenly. At the channels it over-extracts (bitter), in the rest it under-extracts (sour) — both end up together in the cup.
Does a WDT tool really help?
Yes. Even loosening and distributing breaks up clumps and reduces the density differences that cause preferred water paths. Just as important, though, is a level, even tamp.
How hard should I tamp?
Level matters more than hard. In a measurement series with 150 shots at 10, 15 and 20 kg (Kaffeemacher 2025), there was no detectable difference in flow time and TDS. Once the puck is compacted, more force hardly adds anything — a crooked tamp, on the other hand, almost certainly adds a channel.
Do I need a puck screen?
You don't need one, but it helps: it distributes the water more evenly at the top and keeps the shower screen clean. Measurement series on one setup each (Ribes 2021, Clive Coffee 2025) show around 2 to 3 percentage points more yield, mainly because the puck edge extracts more evenly — community measurements, not re-measured by us. Against channeling, distribution in the basket still comes first. In the simulator, the screen lowers the channeling risk by 12 of 100 points, so about as much as 19 points of better distribution; this number is estimated, not measured.
Community measurementS. Ribes, “Effects of a puck screen on extraction” (April 2021), documented at Decent Espresso, link; S. Jemison, Clive Coffee, “Puck Screens: How Much do They Really Impact Extraction?” (17.07.2025), link; both accessed 16.10.2026.
Why does my espresso taste sour and bitter at the same time?
That's the typical sign of channeling: over-extracted channel (bitter) and under-extracted rest (sour) in the same cup. Better distribution and a level tamp usually solve it. Only once the puck flows evenly is it worth adjusting the ratio or grind size — one adjustment at a time.
You'll find all interactive tools bundled in our Coffee Tools.
