TDS and Extraction Yield: Reading the Two Numbers in a Cup

Updated: 7 September 2026

3 key takeaways

• TDS and extraction yield are two distinct percentages: the first is divided by the mass of the beverage, the second by the mass of dry grounds.
• The filter reference window, 1.15 to 1.35% TDS for 18 to 22% yield, comes from preference studies run in the United States in the late 1950s.
• Read together, the two values separate four situations, two of which are fixed by the brew ratio and two by grind or contact time.

Two numbers describe an extraction, and they get confused constantly. TDS says how concentrated the drink is. Extraction yield says how much of the grounds actually made it into the cup. These are not two ways of saying the same thing: one cup at 1.30% TDS corresponds to a yield of 20.8% with 20 g of grounds, and to 16.6% with 25 g for the same amount of beverage. The brew ratio is what makes the difference, and the brew ratio is what links the two quantities. This guide sets out both definitions, the formula, the coordinate system a result is read in, and the limits of that framework. The measuring device itself is covered separately, in the refractometer guide.

What does a TDS of 1.25% actually mean in a cup?

That 100 g of beverage holds 1.25 g of dissolved matter, meaning sugars, acids, emulsified lipids, caffeine and aroma compounds, and 98.75 g of water. The denominator is the mass of the beverage, never the mass of the grounds. That is exactly what makes TDS a concentration rather than a yield.

TDS says nothing about how well the coffee was extracted. It says how dense the drink is. Espresso sits between 8 and 12%, filter coffee around 1.2%, a cold brew concentrate above 2%. Those gaps signal no hierarchy: they reflect different coffee to water ratios. Two drinks with identical TDS may have been extracted very differently, and two identical extractions can produce very different TDS depending on how much water was used.

Hence the first rule of reading: a TDS on its own cannot be interpreted. It becomes useful only alongside the two masses, grounds and beverage, that turn it into a yield.

How much of the ground coffee actually ends up in the cup?

Between 18 and 22% inside the reference window, and never 100%. A yield of 20% means that out of 20 g of ground coffee, 4 g of soluble matter passed into the liquid; the remaining 16 g stays in the bed as cellulose, proteins and fibres that do not dissolve.

The ceiling is a matter of composition rather than technique: roughly 30% of the mass of roasted coffee is water soluble. A yield of 100% has no physical meaning, and even 30% would require dissolving the last fractions, the ones that arrive at the end of an extraction and that nobody is trying to put in a cup. The 18 to 22% window covers the territory where the sought-after fractions have come out without the last ones following.

Here the denominator is the mass of dry grounds. That is what separates yield from TDS completely. Writing "18 to 22% TDS" is an error: the 18 to 22% window belongs to yield, and a filter coffee at 18% TDS would be fifteen times more concentrated than a normal filter coffee, and close to twice an espresso.

How do you calculate extraction yield from a TDS reading?

With one multiplication and one division:

yield (%) = TDS (%) × beverage mass (g) / dry coffee mass (g)

No factor of 100 appears. It shows up only if TDS is entered in decimal form, 0.0130 instead of 1.30, in which case the result has to be multiplied by 100. Both conventions circulate, and crossing them is where yields of 0.2% or 2,080% come from. A plausible yield lives between 15 and 25%.

• Grounds: 20 g
• Beverage, weighed: 320 g
• Measured TDS: 1.30%
• Yield = 1.30 × 320 / 20 = 20.8%

Weigh the beverage rather than deriving it from volume. Treating density as 1 stays acceptable for filter, where dissolved solids account for a little over 1% of the mass, but it breaks down for espresso, where crema also makes the volume unreadable.

Which TDS and yield figures should you aim for by method?

Only one method has a normative window, filter coffee, and that window is the SCA Gold Cup standard. The rest have observed orders of magnitude, not published targets. The yield column makes the point: nothing licenses carrying the 18 to 22% window over to a moka pot or a cold brew, whose extraction mechanics and contact times bear no comparison.

Read this table down its columns rather than across its rows: the TDS column tracks the brew ratio, the yield column does not. That is the shortest demonstration that the two quantities are independent.

How do you get a TDS number you can actually use?

Three inputs feed a yield calculation, and the TDS reading is only one of them. The other two, the mass of grounds and the mass of beverage, cost less to obtain and go wrong just as easily. Choosing an instrument, reading its specification sheet and running its measurement protocol belong to the refractometer guide; what follows keeps only what decides whether the calculation is valid.

The three inputs

• Mass of grounds: weighed dry, before brewing, to a tenth of a gram. Coffee that has picked up ambient moisture weighs more than its true dry mass, which quietly lowers the calculated yield.
• Mass of beverage: weighed after brewing, never inferred from the water poured. This is the most common and most expensive mistake, since it overstates yield by roughly two points in a filter brew.
• The TDS reading: taken on a stirred sample brought down to 20 to 25°C, after zeroing on distilled water. Two or three readings, keeping the median.

Making the number worth something

1. Stir before sampling: a stratified carafe reads differently at the surface and at the bottom, and an unstirred espresso gives a figure with no relation to the drink itself.
2. Filter an espresso sample: suspended fines scatter light and inflate the reading.
3. Log all three masses with the reading: a TDS recorded without its dose and beverage mass cannot be used later, because the yield can no longer be recomputed from it.
4. Change one variable between measurements: two simultaneous changes make the observed difference unattributable.

What does a pair outside the target window tell you?

It tells you which of the two variables to correct, which neither value discloses on its own. Four situations arise, and they call for different moves:

• Low yield, low TDS: extraction is incomplete and the drink is thin. Grind finer, brew hotter or extend contact time, one change at a time.
• Low yield, high TDS: little coffee extracted, but into very little water. Rare in filter, routine in a ristretto. The ratio needs opening up, not the grind tightening.
• High yield, low TDS: plenty extracted, then diluted in too much water. The ratio is at fault, not the extraction: cut the water or raise the dose, and leave the grind alone.
• High yield, high TDS: extraction ran too far in an already dense drink. Grind coarser or shorten contact time.

The dividing rule fits in one sentence: yield is corrected through grind, temperature and time; TDS is corrected through the coffee to water ratio. Two measurements bracketing a single change are enough to know which way you are moving. What measurement does not give you is a sensory diagnosis: reading a cup by taste is covered in the under and over-extraction guide.

Numbers do not validate a cup, they explain why it is what it is. A coffee can stay excellent at 23% yield and turn flat at 20%. Measurement exists to reproduce, not to approve.

← Back to guides

Why do the water-based and beverage-based formulas disagree?

Because the coffee bed keeps part of the water, and only one of the two formulas accounts for it. The mass balance behind extraction yield is simple, but it holds on one condition: the mass entered must be the mass of liquid actually collected, not the mass of water poured over the grounds.

A filter bed retains roughly 2 g of water per gram of dry coffee, with real values running from about 1.5 to 3 g depending on the brewer and the grind. Pour 360 g of water over 20 g of coffee and around 40 g stay in the bed, leaving 320 g of beverage. Run the numbers both ways at a TDS of 1.30%: the beverage form gives 1.30 × 320 / 20 = 20.8%, while the water form, taken naively, gives 1.30 × 360 / 20 = 23.4%. Same brew, two and a half points apart, and the higher figure is the wrong one.

This single omission explains a large share of the improbably high yields circulating in brewing logs. It is also why a water-based calculation can only ever be an estimate: retention is not constant, and every gram of uncertainty on it moves the yield. If you have a scale under the carafe, the question does not arise. If you do not, subtract two grams per gram of dose and treat the result as approximate.

One further caveat applies to both forms. The water retained in the bed carries dissolved solids with it, and those solids never reach the refractometer. What the calculation reports is therefore the yield delivered into the cup, not the total mass dissolved out of the grounds. The two differ slightly, and consistently in the same direction.

Where does the golden cup window come from, and what is it worth?

It comes from work published in 1957. The biochemist E. E. Lockhart, first director of the Coffee Brewing Institute, released The Soluble Solids in Beverage Coffee as an Index to Cup Quality that year and built the Brewing Control Chart: a plane with strength on one axis, extraction yield on the other, and brew ratio drawn as diagonal lines. The central rectangle, 1.15 to 1.35% TDS against 18 to 22% yield, was carried forward by the SCAA and then the SCA as the Gold Cup standard.

That rectangle is not a physical boundary. It is the preference zone of a panel of American consumers in the late 1950s, drinking commercial blends whose roast and freshness bear no relation to today's specialty coffee. Surveys run since in Europe and the Nordic countries record a preference for drinks stronger than the 1.35% upper bound. That does not invalidate the framework; it moves the rectangle, not the axes.

The chart itself has been rebuilt. A team from the UC Davis Coffee Center, working with the SCA and the Coffee Science Foundation, published a new Coffee Brewing Control Chart in the Journal of Food Science in 2023, grounded in trained sensory panels and consumer preference testing. Rather than fencing off an acceptable zone, it maps flavour: bitter and roasted attributes cluster where TDS and yield are both high, sour and fruit attributes where TDS is high and yield low, black tea notes where yield is high and TDS low.

The practical consequence is a change in the window's standing. It remains an excellent starting point for an unfamiliar recipe, and a poor court of appeal for a cup already tasted. A coffee that people enjoy at 23% yield is a coffee that people enjoy at 23% yield, not a mistake awaiting correction.

How much precision can you claim from a calculated yield?

Less than most brewing logs imply, and enough to work with. Yield is not measured, it is computed from three numbers, each carrying its own uncertainty. The TDS reading dominates: manufacturer specifications for dedicated digital instruments sit in the order of ±0.03%. Because yield is the TDS multiplied by the ratio of the two masses, that uncertainty is multiplied by the same factor. In the worked example, 320 g of beverage over 20 g of grounds gives a factor of 16, so ±0.03% of TDS becomes roughly ±0.5 points of yield.

A yield reported as 20.8% should therefore be read as somewhere between 20.3 and 21.3%, before counting weighing error and moisture picked up by the grounds. Two recipes half a point apart are not distinguishable: as far as the measurement is concerned, they are the same. Two points apart is a clear signal.

That imprecision does not undermine the number, it defines what the number is for. Measured the same way every time, with the same instrument and the same sampling routine, yield is an excellent comparison tool: it tells you in which direction and by how much a change of grind or temperature moved the extraction. Presented as the exact quantity of matter dissolved, it claims an accuracy the method cannot supply.

The discipline that follows is unglamorous and decisive. Log the dose, the beverage mass, the TDS and the one variable you changed, in that order, and the series becomes readable after a dozen brews. Log the TDS alone and you have a number that can never be recomputed, compared or explained.