You weigh your dose to a tenth of a gram. You time your shot to the second. You bought a grinder that cost more than your first car. And then you brew with whatever comes out of the tap - which is, by mass, more than ninety percent of what ends up in the cup. Water is the largest ingredient in espresso by an enormous margin, it is the one most people never touch, and it is the last big upgrade available to a setup that already has a decent grinder.
The Ingredient Hiding in Plain Sight
A double shot is roughly 18 g of coffee and 36 g of liquid out. Of that liquid, only about 1.5 to 2.5 g is dissolved coffee solids. Everything else is water and whatever was already dissolved in it before it ever met your beans.
That matters because water is not a neutral carrier. Extraction is a chemical process, and the minerals already in the water are active participants in it. Change the water and you change the extraction itself - not just the background flavour, but how much coffee actually comes out of the puck at the same grind, dose and time.
Forget TDS. Learn Two Numbers Instead.
Most water advice starts and ends with TDS - total dissolved solids, the number a cheap pen meter gives you. It is the least useful of the common figures, and leaning on it is the most common mistake in this whole subject.
A TDS pen does not measure dissolved solids at all. It measures electrical conductivity and converts it, using a fixed assumption, into an estimated ppm figure. It cannot tell you which solids are in there. Water at 150 ppm that is mostly sodium behaves nothing like water at 150 ppm that is mostly calcium and bicarbonate, but the pen reports the same number for both. As Scott Rao puts it, TDS by itself is a relatively unimportant number.
The two numbers that actually govern what happens in your cup and inside your boiler are these:
Both are conventionally reported in ppm (or mg/L) "as CaCO₃", an equivalence convention that lets you compare calcium, magnesium and bicarbonate on a single scale. When a water recipe says "50 ppm", it almost always means 50 ppm as CaCO₃.
Hardness: Why Magnesium Extracts Harder Than Calcium
Calcium and magnesium ions carry a positive charge, and many of the flavour-carrying compounds in coffee carry a negative one. The cations bind to those compounds and help carry them into solution. This is not folklore - it was modelled and published by Hendon and colleagues in 2014, and their headline finding was that magnesium is the more effective extractor of the two, with calcium somewhat behind it and sodium contributing very little.
This is why nearly every DIY water recipe is built on magnesium sulfate - Epsom salt - rather than a calcium salt. Magnesium extracts more per unit of hardness, and unlike calcium it is far less prone to depositing as boiler scale.
Alkalinity: The Number That Quietly Flattens Your Coffee
Alkalinity is the water's capacity to neutralise acid. Coffee is full of acids - citric, malic, quinic, chlorogenic and others - and those acids are a large part of what makes a good light roast interesting. Bicarbonate in your water neutralises them on contact.
Rao calls alkalinity by far the single most important factor in how water chemistry affects coffee flavour, and the relationship is blunt: higher alkalinity gives a less acidic cup, lower alkalinity a more acidic one.
High alkalinity is the reason so much home espresso tastes muted. It does not taste obviously wrong. It tastes like a slightly boring version of the coffee you paid for: the fruit is gone, the brightness is gone, the sweetness reads as flat and chalky, and everything trends toward generic roast. People chase this with the grinder and the temperature for months without ever suspecting the tap.
The Water Map
Because one number drives extraction and the other suppresses acidity, you can place almost any water on a simple grid and predict the cup.
| Low alkalinity (KH under ~40) | High alkalinity (KH over ~80) | |
|---|---|---|
| Low hardness (GH under ~40) | Thin but bright. Sharp, sour, hollow. Acidity with nothing underneath it. Common with RO and over-filtered water. | The dead zone. Flat, papery, weak and dull all at once. Nothing to extract with, and nothing left to taste. |
| High hardness (GH over ~100) | Big, intense, sometimes harsh. Strong extraction with no brake - can read as aggressive or astringent, but rarely boring. | Heavy and muddy. Strong but one-dimensional, chalky, mineral-tasting. Classic hard-tap-water espresso, and the worst case for scale. |
The interesting territory is the middle: enough hardness to extract properly, little enough alkalinity to let the coffee's own acidity survive.
What to Actually Aim For
There is no single correct water, and anyone who tells you otherwise is selling something. There are, however, three well-established reference points.
| Target | Hardness (GH) | Alkalinity (KH) | Best for |
|---|---|---|---|
| SCA standard | 68 ppm (range 17-85) | around 40 ppm | The safe general-purpose reference, filter and espresso alike. TDS target 150 ppm, acceptable 75-250. |
| Espresso-practical | 50-80 ppm | 30-50 ppm | Where most home espresso recipes land. Good extraction, low scale risk, acidity intact. |
| Low-buffer (Rao style) | 50-80 ppm | 30-40 ppm | Light, modern, fruit-forward roasts where you want maximum acidity and clarity. |
| Avoid | over ~100 ppm | over ~80 ppm | Scale risk climbs steeply and the cup flattens. Also avoid near-zero on both - see corrosion below. |
If you want one line to carry away: aim for roughly 50-80 ppm hardness and 30-50 ppm alkalinity, and keep hardness comfortably above alkalinity. Everything else is refinement.
Reading a Bottled Water Label
You do not need a lab. Almost every bottled water prints its mineral analysis on the side, and three lines on that label tell you what you need. The figures are given in mg/L of the raw ion, so you convert them to the ppm-as-CaCO₃ scale that coffee recipes use.
Alkalinity (KH) = bicarbonate (HCO₃) × 0.82
Both results come out in ppm as CaCO₃. If the label shows "dry residue at 180 °C", that is roughly the TDS figure - useful as a sanity check, not as a target.
A worked example. Say a label reads calcium 11.5 mg/L, magnesium 8.0 mg/L, bicarbonate 71 mg/L.
- Hardness = (11.5 × 2.5) + (8.0 × 4.1) = 28.75 + 32.8 = about 62 ppm
- Alkalinity = 71 × 0.82 = about 58 ppm
That is a decent espresso water: hardness in range, alkalinity a little higher than ideal, so expect a slightly rounded, less acidic cup. Perfectly usable, and vastly better than hard tap water.
Run the same arithmetic on three or four bottles at your supermarket and you will usually find one that lands close to target. That is the cheapest possible upgrade, and in many countries it costs less than a filter cartridge.
Unit conversions, because every country uses a different one
| Unit | Used in | Equals in ppm CaCO₃ |
|---|---|---|
| 1 °dH (German degree) | Germany, Austria, central Europe | 17.8 ppm |
| 1 °fH (French degree) | France, Italy, Spain | 10 ppm |
| 1 gpg (grain per gallon) | United States | 17.1 ppm |
| 1 mmol/L | Scientific reports | 100 ppm |
So a German water report listing 12 °dH is about 214 ppm hardness - roughly three times what you want for espresso, and firmly in scale territory.
The Machine Side: Two Ways Water Destroys Equipment
Taste is only half the argument. The other half is that water decides how long your machine lives, and it can kill it from either direction.
Too hard: scale
Calcium carbonate is one of the few compounds that becomes less soluble as water gets hotter. Heat it in a boiler and it comes out of solution and plates itself onto the hottest surfaces - the heating element, the boiler walls, the narrow passages of a thermoblock.
The damage is progressive and mostly invisible. Scale insulates the heating element, so the machine runs hotter internally to hit the same setpoint and temperature stability degrades. It narrows passages, so flow drops and your brew ratio quietly shifts. In thermoblock machines - most entry-level units and many bean-to-cup machines - the passages are so narrow that scale can block them permanently, and the part is often not descalable in practice.
Too pure: corrosion
The reflex fix - "I will just use distilled water" - creates the opposite problem. Water with almost no dissolved minerals is chemically aggressive. It is under-saturated, so it dissolves what it touches: copper, brass, the solder in the joints. Over time this produces pitting corrosion, metal leaching, and metallic off-flavours in the cup.
A modest amount of bicarbonate is what keeps water from behaving this way, which is one of several reasons that zero-hardness recipes are more debated than they first appear. Note also that chloride - a different thing from the chlorine used to disinfect tap water - attacks stainless steel specifically; a commonly cited safe ceiling is somewhere around 15 to 30 ppm.
Your Five Options, Ranked by Effort
| Approach | Effort | Taste | Machine protection | Verdict |
|---|---|---|---|---|
| Tap plus carbon filter (jug filter) | Minimal | Removes chlorine; barely touches hardness or alkalinity | Little to none | Fine only if your tap water is already soft. Check your utility's report first - it is usually published online. |
| Bottled water | Low | Excellent if you pick by label | Excellent | The best value-for-effort option for most people. Read the label with the formulas above. |
| Ion-exchange softener cartridge | Low | Often worse - see warning | Good against scale | Protects the machine, can flatten the cup. Understand what it does before buying. |
| Dealkalising cartridge (cafe-style, with bypass) | Medium | Very good, tunable | Excellent | What cafes use. The bypass dial blends untreated water back in, so you can tune hardness. |
| Distilled or RO plus a mineral recipe | Medium-high | Total control, fully repeatable | Excellent | The enthusiast answer. Identical water every single time, anywhere in the world. |
Building Your Own Water
Distilled or RO water plus two salts gives you a water you can reproduce exactly, forever, regardless of what the utility does this month. You need magnesium sulfate heptahydrate (Epsom salt, food or USP grade) and sodium bicarbonate (plain baking soda). Potassium bicarbonate can replace the baking soda if you would rather not add sodium.
Method A: direct dosing, per gallon (3.78 L)
| Recipe | Epsom salt | Bicarbonate | Roughly GH / KH | Character |
|---|---|---|---|---|
| Barista Hustle #4 | 0.745 g | 0.254 g baking soda | 80 / 40 | Balanced, close to the SCA reference. A safe default. |
| Third Wave Water, espresso profile | 1.489 g | 0.510 g potassium bicarbonate | 160 / 68 | Higher extraction and body. Strong, and forgiving on dark roasts. |
| RPavlis | none | 0.378 g potassium bicarbonate | 0 / 50 | Zero hardness, popular because it cannot scale - but the flattest of the three, and worth reading the corrosion debate before committing. |
Method B: concentrates, which is what you will actually do
Weighing 0.25 g accurately every time is miserable. Instead, make two strong stock solutions once and dose them by volume.
- Concentrate A (hardness): dissolve 25 g of Epsom salt into 1000 g of distilled water.
- Concentrate B (buffer): dissolve 8.6 g of baking soda into 1000 g of distilled water.
- To use: add a small measured amount of each - for a balanced profile, on the order of 10 mL of A and 10 mL of B per 4 L of distilled water - then shake and use. Scale the two independently: more A for body and extraction, less B for more acidity.
How to Measure What You Have
Skip the TDS pen as a primary tool. The cheapest genuinely useful kit is an aquarium GH/KH drop test - the kind that costs about the price of two bags of coffee and lasts for years. You count drops until the colour flips, each drop is one degree, and you multiply by 17.8 for ppm as CaCO₃.
- Free and instant: your water utility publishes an annual quality report with hardness and alkalinity for your area. Start there.
- Cheap and accurate enough: an aquarium GH/KH drop kit. This is the right answer for almost everyone.
- Visual evidence: look inside your kettle. A chalky white crust means high hardness and high alkalinity, and your espresso machine is getting the same treatment.
- TDS pen: useful for one job - confirming your RO or distilled water really is near zero before you mineralise it.
Diagnosing Water From the Cup
Before you buy anything, run one controlled test. Pull your normal shot with your normal water. Then pull the identical shot - same dose, same grind, same time - with a suitable bottled water. Change nothing else. The difference is your water's signature. Pulling both shots with the same step-by-step workflow is what makes the comparison fair.
One caution: water problems look a lot like several other problems. If the cup is sour and thin, confirm it is not simply under-extraction from a coarse grind setting, and if it is bitter and harsh, rule out temperature first. Work through the standard troubleshooting order before blaming the tap, and use the dial-in tool to keep every other variable fixed while you test.
What to Do This Week
- Find out what you have. Look up your utility's hardness and alkalinity figures, or spend ten minutes with a GH/KH drop kit.
- If hardness is over about 100 ppm, change something today. Not for taste - for the machine. Scale is cumulative, and it is the most expensive mistake on this page.
- Run the two-shot comparison. Your tap water against a bottled water you have checked with the formulas above, with everything else held identical.
- Pick the lowest-effort option that gets you into range. For most people that is a bottled water chosen by its label, not a mineral kit.
- Re-dial once, then leave it alone. New water changes extraction, so expect your grind setting to move. Fix the water first, then dial the grind - not the other way around.