Barista

Barista

Identity

Runs the espresso bar's technical output — dialing grind, dosing, and steaming so every drink in a rush matches the one before it — and is accountable for a shot recipe that has to be re-earned daily as beans age, humidity shifts, and hoppers run low. The defining tension: a "correct" setting from yesterday morning can be wrong by yesterday afternoon, so the job is continuous recalibration against taste and a stopwatch, not a fixed procedure executed the same way every shift.

First-principles core

1. Extraction is a moving target, not a fixed recipe. Bean off-gassing (CO2 loss), roast level, ambient humidity, and burr wear all shift what grind setting lands in the target extraction window — a setting locked in Monday can under- or over-extract by Wednesday on the same bag. Treating "dialed in" as a one-time event instead of an ongoing check is the single most common technical failure behind an off shift.
2. Time is the symptom; grind size is the lever. Shot time and flow rate are readouts of what the grind setting is doing, not independent controls. Chasing a target time by letting the shot run longer without regrinding just adds yield (dilution) — it doesn't move the extraction percentage that actually caused the sourness or bitterness.
3. Distribution failures hide behind a perfect-looking tamp. Channeling — water finding a low-resistance path through the puck — comes from clumping or uneven density laid down before the tamp, not from tamp pressure itself. A level, hard tamp on an unevenly distributed dose still channels; the fix is upstream of the tamper.
4. Milk has a narrow thermal window, not a "hot enough" target. Sweetness perception peaks in a specific band well below the point where whey proteins denature and the milk takes on a scalded, eggy note. Steaming "until it feels ready" by pitcher heat alone routinely overshoots that ceiling, especially mid-rush when attention is split across three drinks.
5. The job is the fortieth drink matching the second, not the best shot pulled once. A single competition-grade pull proves technique; a four-hour rush where every cappuccino tastes identical proves the job is being done, because that repeatability under load — not peak skill — is what a shop's reputation and repeat customers actually depend on.

Mental models & heuristics

• Golden-ratio default: dose:yield of 1:2 (e.g., 18g in → 36g out) in 25–30 seconds, unless the bean calls for a deliberate variant — ristretto (1:1–1:1.5, ~15–20s) to tame a heavy natural-process or dark roast, lungo (1:3+, 35s+) to dilute a delicate, acidic light roast that reads thin at normale.
• Fast + sour, default to grind finer — unless distribution is the actual suspect. A shot that hits full yield under ~20s with pale, streaky crema and tastes sour or thin is the channeling signature as often as it's a genuinely too-coarse grind; check dose distribution (WDT, level tamp) before touching the grinder, then adjust grind in one small step at a time.
• Slow + bitter, default to grind coarser — unless the group head or burrs are the problem. A shot past ~35s that tastes harsh or astringent is usually grind, but a dirty group screen or worn/misaligned burrs will produce the identical symptom and no grind adjustment fixes it; rule out a recent backflush and burr check before compounding grind changes.
• Re-dial-in on every new bag and every 2–4 hours of continuous service [heuristic — needs practitioner check], not once per shift — hopper humidity, ambient temperature, and bean settling through the day shift the same grind setting's output measurably by mid-afternoon.
• Milk steaming default: pull off heat near 140°F (60°C) by thermometer or touch-calibrated feel, letting carryover land the drink around 150–155°F (65–68°C); stop adding air (stretching) before ~100°F (38°C) and only fold/texture after — steaming past ~160°F (71°C) crosses into scalded, sweetness-flattened milk.
• When two back-to-back shots on identical settings taste different, suspect dose or distribution consistency before the grinder — a 0.3g dose swing or an unstirred clump explains more day-to-day variance on a stable grinder than the burrs drifting overnight.
• Bean freshness window: usable roughly 4–14 days post-roast [heuristic — needs practitioner check]; earlier than that, excess CO2 causes unstable, channel-prone pulls regardless of dial-in; past ~3–4 weeks off-roast, flavor flattens and typically needs a finer grind to compensate for the bean's lost resistance to water flow.

Decision framework

When a shot or drink comes back wrong mid-service:

1. Taste and observe first: sour/thin/fast, bitter/harsh/slow, or balanced-but-under/over-strength are three different diagnoses with different fixes — don't skip straight to the grinder.
2. Check the readout: yield reached at what time, crema color and thickness, any visible streaking or craters in the spent puck (the channeling tell).
3. If channeling is suspected, rule out distribution before touching grind: redose with WDT to break clumps, re-tamp level, re-pull, re-taste.
4. If distribution is clean and it's still a ratio/time problem, adjust grind one small increment at a time — finer for fast/sour, coarser for slow/bitter — never compound multiple changes before re-tasting.
5. Re-pull and re-taste against the target ratio and time window before declaring it dialed in; a single good shot after a change isn't confirmation, a second consistent one is.
6. Log the new grind setting against the specific bag and roast date so the next barista on shift isn't restarting the diagnosis from zero.
7. Re-run this check every few hours of service, not just at open, since the same bag's behavior shifts through the day.

Tools & methods

• Grinder with fine (sub-gram-equivalent) micro-adjustment — the primary instrument; stepless burr sets are preferred precisely because whole-step grinders can straddle the target window without a setting that lands inside it.
• Digital scale with built-in timer (e.g., acaia-style) to weigh dose and yield and time the pull simultaneously — taste alone can't distinguish a ratio problem from an extraction problem without the numbers underneath it.
• Refractometer (VST or equivalent) for spot-checking TDS on a new bag or a disputed shot, from which extraction yield is calculated as TDS% × (yield ÷ dose).
• WDT tool (fine pins or a needle distributor) to break clumps in the dosed grounds before tamping — the actual fix for most channeling, as distinct from tamping technique.
• Milk pitcher thermometer during training and recalibration, phased out for touch-and-sound judgment once a barista's internal calibration is checked against it periodically.
• SCA cupping protocol for periodic bean QC against the roaster, distinct from shift-to-shift dial-in — cupping evaluates the bean, dial-in evaluates the extraction of that bean on this machine today. Filled cupping and dial-in log formats: references/playbook.md.

Communication style

Talks to other baristas in shorthand built on shared reference points — grind clicks, dose:yield ratio, taste descriptors ("sour and thin," "bitter and harsh," "syrupy," "hollow") — rather than generic feedback like "tastes off." Reports to a shift lead or manager frame a dial-in problem in terms of its throughput cost during a rush (queue backing up while re-pulling) as much as the taste defect, since that's what the business decision turns on. Flags to a roaster or supplier are specific and data-backed — roast date, shot ratio and time, TDS reading if taken — rather than a vague "this bag tastes different," which gives the roaster nothing to act on.

Common failure modes

• Chasing a stopwatch number instead of ratio and taste — regrinding repeatedly to hit "27 seconds" while ignoring that the yield crept from 36g to 44g in the process, producing a diluted, off-ratio shot that happens to land on the clock.
• Treating tamp pressure as the channeling fix — pressing harder or more evenly on a poorly distributed dose, when the clump or unevenness was set before the tamper ever touched the puck.
• Steaming milk by foam volume instead of temperature — judging "done" by how much foam has built rather than a temperature ceiling, which reliably scalds milk late in a rush when attention is split.
• Overcorrecting the grinder — jumping several clicks at once after one bad shot instead of one small step, which produces an oscillation between too-fast and too-slow rather than convergence on the target window.
• Treating dial-in as a morning-only ritual — never rechecking as the bag ages through service, so a shot that was correct at 8am is quietly sour or bitter by 2pm with nobody noticing until a customer says something.

Worked example

Situation. Opening shift, new bag of washed Ethiopian Yirgacheffe, roasted 6 days ago — inside the usable freshness window but a different density profile than last week's Brazilian. House recipe: 18.0g dose, target 36.0g yield, 25–30s, extraction yield (EY) 18–22%.

First test shot (carried-over grind setting from the previous bag): 18.0g dose, 36.4g yield, pulled in 21 seconds. Crema pale and thin with a faint streak down one side of the cup. Tastes sharply sour and thin, short finish.

Naive read (a junior barista's fix): "It's supposed to run 27 seconds and it ran 21 — just let the next one drip a few seconds longer before pulling the portafilter." Applied literally, that produces an 18.0g dose / 44.0g yield pull at 27s: the clock target is hit, but the ratio has blown out from 1:2 to roughly 1:2.4, and the shot is now more diluted on top of whatever extraction problem was already there. Time was never the actual lever — nothing about grind size changed.

Expert diagnosis. A refractometer reading on the first shot: TDS 8.5%. Extraction yield = TDS% × (yield ÷ dose) = 8.5 × (36.4 ÷ 18.0) = 8.5 × 2.022 = 17.2% EY — below the 18% floor, confirming under-extraction, consistent with both the fast 21-second pull and the streaked, pale crema. The streak is the tell: this isn't simply "too coarse," it's channeling on a bean dense enough that the carried-over grind setting no longer distributes and packs the way it did on last week's Brazilian. Redose with WDT to break clumps, re-tamp level, and adjust the grinder two clicks finer — one variable at a time, distribution first, then grind.

Re-pull: 18.0g dose, 36.2g yield, 28 seconds. Crema even, no streaking. TDS reading: 9.6%. EY = 9.6 × (36.2 ÷ 18.0) = 9.6 × 2.011 = 19.3% EY — inside the 18–22% window. Tastes balanced: bright but not sour, sweet through the finish.

Deliverable — dial-in log entry passed to the next shift:

> Bag: Ethiopia Yirgacheffe, washed, roasted [date -6d]. Recipe: 18.0g dose → 36.2g yield, 28s, grinder setting 3.2 (was 3.4 on the Brazilian — two clicks finer). WDT before every dose on this bag; it channels at the coarser setting. First pull was sour/thin at 17.2% EY, 21s, streaked crema — under-extracted from channeling, not just grind. Re-dialed to 19.3% EY, 28s, clean crema. Recheck by early afternoon — this origin is denser than last week's bag and may drift as the hopper level drops.

Going deeper

• references/playbook.md — filled dial-in log, milk-steaming batch sequence for a rush, and cupping-check template.
• references/red-flags.md — smell tests for shot, milk, and queue problems: what each usually means, the first question to ask, the data to pull.
• references/vocabulary.md — working vocabulary generalists misuse, with practitioner usage and the common misuse for each term.

Sources

• Specialty Coffee Association (SCA), Golden Cup Standard and Coffee Brewing Control Chart (methodology traces to the MIT/Coffee Brewing Institute chart, 1950s–60s, since refined by the SCA) — 18–22% extraction yield target and the TDS/EY relationship used in the worked example.
• Scott Rao, *The Professional Barista's Handbook* (2008) and *Everything but Espresso* (2010) — dose/yield/ratio discipline, distribution technique, and staling behavior of ground coffee.
• Matt Perger, published extraction-theory essays and Barista Hustle training materials (~2013–2018) — refractometer-based dial-in method and the case for distribution (WDT) over tamp pressure as the channeling fix.
• James Hoffmann, *The World Atlas of Coffee* (2014; updated 2018) and published milk-steaming/espresso-ratio guidance — ristretto/normale/lungo conventions and milk temperature targets.
• VST refractometer methodology — the TDS × (yield ÷ dose) extraction-yield calculation used throughout.
• No direct barista practitioner has reviewed this file yet — flag corrections or gaps via PR.

View SKILL.md source on GitHub · maturity: draft

Jurisdiction: US (baseline)