How a South Indian coffee filter extracts its strong decoction

The steel filter is a slow gravity percolator with a high coffee-to-water ratio. Run the mass balance and the decoction comes out several times stronger than a cup of drip coffee, before any milk goes in.

Steel tumblers of frothy South Indian filter coffee on a table at Mavalli Tiffin Room in Bengaluru
Photo: Charles Haynes, CC BY-SA 2.0, via Wikimedia Commons

A South Indian coffee filter makes its strong decoction by slow gravity percolation: hot water sits on a packed bed of finely ground, dark-roasted coffee (often blended with roasted chicory) and drips through a perforated plate over 10 to 20 minutes. The trick is the dose, not the device. With roughly one part powder to five parts water, a normal extraction of about 20% gives a decoction near 5 to 7% dissolved solids, against the 1.15 to 1.35% that the Specialty Coffee Association calls ideal for a cup of drip coffee.

Key takeaways
  • 5–7% TDS is my mass-balance estimate for a kaapi decoction, about 5 times a drip cup.
  • 1:5 powder to water, roughly, is what makes it strong; the device is just a slow percolator.
  • Grind and dose matter more than water temperature between 87 and 93°C.
  • 51% coffee minimum: chicory can be up to 49% of an FSSAI-legal blend.
  • The tumbler-dabara pour cools the coffee and whips air into the milk.

That number explains everything about kaapi. The decoction is a concentrate, built to be cut with two or three parts of hot milk and still taste of coffee.

Below I work through the filter as a piece of process engineering:

  • the mass balance that sets strength;
  • what each variable does;
  • what chicory changes chemically;
  • why pouring between a tumbler and a dabara is a cooling and aeration step rather than theatre.

The numbers come from the SCA's brewing standard, a 2020 sensory study from UC Davis, a 2016 NMR study of roasting chicory root, FSSAI rules on coffee-chicory blends and the Coffee Board of India's latest statistics.

1.15–1.35%SCA "Golden Cup" strength for brewed coffee (TDS)
18–22%SCA target extraction yield
51%minimum coffee share in an FSSAI coffee-chicory mixture
96,000 tIndia's domestic coffee use in 2024 (Coffee Board, CRISIL estimate)

How does a South Indian coffee filter work?

The filter is a two-tier steel cylinder: an upper cup with a finely perforated floor that holds the powder, a pressing disc (the "umbrella") that sits on the grounds, a lid, and a lower cup that collects the decoction. Boiling water is poured over the disc, the lid goes on, and gravity does the rest.

In engineering terms it is a batch percolator with no pump and no pressure head beyond a few centimetres of water. Compare that with espresso at about 9 bar, or a pour-over where the brewer controls flow.

In the filter, three things set the flow rate:

  • the size of the holes in the perforated floor;
  • the fineness of the grind;
  • how hard the powder was tamped under the disc.

The water spends a long time in contact with the bed, which is what you need when the dose is high and the bed is deep.

Hot water (about 90–100°C)Pressing disc ("umbrella")Packed bed: fine coffee + chicoryPerforated floorSlow drip, 10–20 minDecoction: 5–7% TDS (estimate) Upper cupLower cup
The filter is a gravity percolator. Grind, tamping and hole size set the drip time; the dose-to-water ratio sets the strength.

Why is filter coffee decoction so strong?

Because strength is set by the ratio of dissolved coffee to water, and the filter uses far less water per gram of coffee than any cup brewer. Extraction yield is similar to ordinary brewing; the concentration is not.

The story

In 1957 E.E. Lockhart, publishing through the Coffee Brewing Institute in New York, tied the quality of a cup of coffee to two numbers: brew strength, the share of dissolved solids in the liquid, and extraction yield, the share of the grounds that dissolved. He drew them on one grid, the Coffee Brewing Control Chart, split into nine regions labelled strong or weak, bitter or underdeveloped. The centre box, 1.15 to 1.35% strength and 18 to 22% extraction, was marked "ideal" and became the SCA's Golden Cup. In 2020 a UC Davis team pointed out that nobody had shown which drinkers found that centre ideal, and tested it with 270 brews. A kaapi decoction sits far off the right-hand edge of Lockhart's grid, on purpose.

Batali, Ristenpart and Guinard, Scientific Reports (2020): introduction on Lockhart's 1957 control chart

The Specialty Coffee Association's brewing standard defines two numbers:

  • Strength is total dissolved solids (TDS), in grams per 100 g of liquid, and its "Golden Cup" range for brewed coffee is 1.15 to 1.35%.
  • Extraction yield is the percentage of the dry grounds that ends up dissolved, with a target of 18 to 22%.

The two are linked by the brew ratio and by how much water stays trapped in the wet grounds. A drip brewer uses roughly 55 to 60 g of coffee per litre. Take a worked example of 20 g of powder and 100 ml of water, close to 200 g per litre.

Calculated in this post: 20 g ÷ 0.1 l = 200 g per litre. The 20 g and 100 ml are the example’s assumed inputs, not a measured home average; the 1.15 to 1.35% and 18 to 22% ranges come from Batali et al. (2020).

Here is the mass balance I use, with the assumptions marked:

  1. Dissolved solids. Dose × extraction yield. At 20 g and 20%, that is 4 g of coffee solids in solution.
  2. Water lost to the bed. Spent grounds hold water. A common brewing estimate is about 2 g of water per gram of grounds, so 20 g of powder keeps about 40 g. This is my working assumption, not a measured figure for kaapi powder, and chicory may hold more.
  3. Decoction mass. 100 g of water − 40 g retained + 4 g dissolved ≈ 64 g.
  4. Strength. 4 ÷ 64 ≈ 6.3% TDS, calculated from the steps above. Change extraction to 16% or 24% and the answer moves to about 5.1% or 7.4%.
  5. In the cup. Add 120 ml of milk to 60 g of decoction and the coffee solids drop to about 2% of the drink (calculated from the steps above), still above the 1.15 to 1.35% of a black drip cup, now carried by milk fat and lactose.

So a decoction is about five times as concentrated as a Golden Cup coffee. It is the same idea as an espresso shot, reached through a very different machine.

Where kaapi sits on Lockhart's chart 02468Strength, TDS (%) 14182226 Extraction yield (%) Golden Cup milky kaapi, about 2% Decoction5.1–7.4% espresso: 8–12% Decoction band: the mass balance above at 16–24% extraction (estimate)
The Golden Cup box from the SCA standard, as described by Batali et al. (2020); the decoction and milky-cup points are the estimates worked out above.

Estimate your filter decoction strength

What sets the extraction: grind, dose, time or temperature?

In a gravity filter, grind and dose do most of the work, and temperature matters less than most people think. Contact time is a result of the grind and the tamp, not a dial you set.

VariableWhat it changes in the filterDirection
Dose (powder per ml of water)Strength of the decoction, almost directlyMore powder, stronger decoction, less of it
Grind sizeFlow rate and extraction yieldFiner: slower drip, higher extraction, risk of a stalled filter
Tamping under the discBed permeability and channellingToo loose: water rushes through gaps; too tight: drip can take an hour
Chicory shareBody, colour, flow and the share of caffeine-free solidsMore chicory: darker, thicker, more bitter-sweet, less caffeine per gram
Water temperatureRate of dissolutionSmaller effect than people assume within 87 to 93°C (Batali et al., 2020)
Roast degreeSolubility and flavourDarker roasts dissolve faster and taste more bitter-roasty

Why temperature matters less than you think

The temperature point deserves attention because the filter cannot hold its heat. Water poured at a boil cools as it sits in the upper cup for ten minutes or more.

A 2020 UC Davis study tested exactly this question for drip coffee and found that, once strength and extraction were matched, brews made at 87, 90 and 93°C tasted almost the same. What changed the taste was TDS and extraction yield.

Brew temperature, at fixed brew strength and extraction, has little impact on the sensory profile of drip brew coffeeBatali, Ristenpart and Guinard · Scientific Reports · 2020

Twelve trained panellists rated 31 attributes in drip coffees brewed at 87, 90 and 93°C, with grind and time adjusted to hit nine target points on the brewing control chart (about 1.0 to 1.5% TDS and 16 to 24% extraction). Across 270 brews, temperature had no significant overall effect, while TDS and extraction yield both did; only "nutty" moved with temperature, by about one point on a 100-point scale. Sourness varied by roughly 20 points between high-strength, low-extraction cups and the reverse. The limit: drip brewing at normal strength, not a 5% concentrate, so it supports the principle rather than measuring kaapi directly.

Read the paper (free)
A steel tumbler and dabara of South Indian filter kaapi with a frothy top
Tumbler and dabara. The decoction is a concentrate; the milk and the pour finish the drink.Photo: Nizil Shah, CC BY-SA 4.0, via Wikimedia Commons

What does chicory do to the decoction?

Roasted chicory root adds dark colour, body and a bitter-sweet, caramel note, and it slows the drip. It adds no caffeine, so every gram of chicory in the blend lowers the caffeine in the cup.

Raw chicory root is rich in inulin, a chain of fructose units. Roasting breaks it down. A 2016 study from the University of Tokyo followed 36 compounds through roasting with NMR and found that inulin, sucrose and most amino acids were largely degraded, while new compounds appeared: 5-hydroxymethylfurfural (HMF), di-D-fructose dianhydrides, acetamide and norfuraneol. Free monosaccharides first fell, then rose again up to the dark-roast stage. Fructose fragments and their caramelisation products are why roasted chicory dissolves readily and colours water so fast.

That chemistry has a practical side. Chicory is mostly carbohydrate, so a blend gives more dark, soluble matter per gram than coffee alone, which makes a decoction look and feel stronger than its coffee content suggests. Many households also find a chicory blend drips more slowly through the same filter. I could not find a peer-reviewed measurement of decoction TDS or flow with and without chicory, so treat the flow effect as kitchen experience rather than a measured fact.

Use of NMR-Based Metabolomics To Chemically Characterize the Roasting Process of Chicory RootWei, Furihata, Zhang, Miyakawa and Tanokura · Journal of Agricultural and Food Chemistry · 2016

Dried chicory root pulp was roasted in a direct-fired roaster to several degrees, in triplicate, and 36 NMR-visible components were tracked. Inulins, sucrose and most amino acids were largely degraded; acetamide, HMF, di-D-fructose dianhydride and norfuraneol formed during roasting. A principal component analysis showed that a hotter, shorter roast and a cooler, longer roast reached similar composition. The limit: Japanese-grown chicory and a lab roaster, and NMR sees only compounds above its detection limit, not the large melanoidin polymers.

Read the paper (free copy)
Cross-section of a chicory root showing its pale, fleshy interior
Chicory root stores inulin, a fructose chain. Roasting breaks it into soluble sugars, HMF and caramel-like compounds.Photo: Sash 2002, CC BY 4.0, via Wikimedia Commons

How much chicory is allowed in Indian filter coffee powder?

Under FSSAI rules a coffee-chicory mixture must contain at least 51% coffee by mass, so chicory can be up to 49%. The pack must state both percentages.

The standard sits in sub-regulation 2.10.4 of the Food Safety and Standards (Food Products Standards and Food Additives) Regulations, 2011. A 2018 FSSAI draft proposed raising the coffee minimum to 70%; I found no notification that made it final. The labelling side moved in 2025 and 2026: FSSAI notified a boxed front-of-pack declaration in August 2025, then in July 2026 allowed a simpler "COFFEE …%, CHICORY …%" text line on the principal display panel and pushed compliance to 1 July 2027.

Common blend labelCoffee shareChicory shareLegal under FSSAI?
Pure coffee100%0%Yes, sold as coffee
90:1090%10%Yes
80:20 (the classic)80%20%Yes
70:3070%30%Yes
53:4753%47%Yes, near the floor
50:5050%50%No, below the 51% coffee minimum

The full trade-offs of chicory versus pure coffee have their own post: should filter coffee have chicory?

Which beans go into filter coffee powder?

Traditional filter blends use dark-roasted Indian coffee, often a washed arabica (the "Plantation" grades) or peaberry, with robusta in many commercial powders for body and price. Robusta dominates what India grows.

The Coffee Board's final figures for 2025-26 put India's crop at 3,73,000 tonnes: 1,11,000 t of arabica and 2,62,000 t of robusta, so robusta is about 70% of output. Its first estimate for 2026-27 is 4,04,000 t. Robusta brings roughly double the caffeine and more chlorogenic acids, which suits a strong, milk-cut drink; the chemistry is in arabica vs robusta in India. Domestic use is small next to output: the Board's CRISIL-based estimate is about 96,000 t in 2024, while most of the crop is exported.

India coffee production by species
See the data as a table
Arabica (tonnes)Robusta (tonnes)
2023-24101500259000
2025-26 (final)111000262000
2026-27 (post-blossom estimate)120000284000

Source: Coffee Board of India: Database on Coffee (July 2024) and Coffee Statistics page · 2024-25 omitted because its final split was not on the pages I used.

Why pour coffee between a tumbler and a dabara?

The long pour cools the coffee fast and whips air into the milk. It turns a scalding mix into a drinkable one in seconds and builds the foam that marks a good kaapi.

Two physical effects are at work:

  • Cooling. A thin, falling stream has a large surface area, so it loses heat by evaporation and convection much faster than a still cup.
  • Foam. The stream entrains air when it hits the liquid below, which hot milk proteins stabilise as foam.

Restaurants pour from a height for both reasons. The dabara (the wide, lipped bowl) gives the stream a broad target and a shallow pool, which cools fastest. I have not found a controlled study that measures the temperature drop per pour, so I give the mechanism, not a number.

How do you fix a weak or slow filter?

Weak decoction almost always means too little powder, too coarse a grind or a loose bed that let water channel through. A filter that won't drip means the grind is too fine or the disc was pressed too hard.

  1. Fix the ratio first. Weigh the powder and the water once. If you are below about 1 g of powder per 6 ml of water, strength will suffer whatever else you do.
  2. Match grind to filter. Filter-grade powder is fine but not espresso-fine. If the first drops appear in seconds, go finer; if it barely drips, go coarser.
  3. Level and tamp gently. Shake the upper cup to level the bed, set the disc and press lightly. Channels form along a lopsided bed.
  4. Pour in two stages. A small first pour wets the bed so it swells evenly; the rest follows a minute later. This is the same "bloom" idea pour-over brewers use.
  5. Don't over-collect. The last drips are the most dilute. Pouring a second round of water through spent grounds gives a thin, bitter "second decoction".

The full routine, with timings, is in how to make South Indian filter coffee at home. For what a decoction does to your caffeine count, see caffeine in Indian drinks ranked.

A cup of South Indian filter coffee with milk froth served in a steel tumbler
A milky kaapi is about one part decoction to two or three parts hot milk.Photo: Babithajcosta, CC BY-SA 4.0, via Wikimedia Commons

Is a filter decoction like espresso?

In strength, yes; in mechanism, no. Both are concentrates near 5 to 10% TDS, but espresso is pushed by about 9 bar in under 30 seconds, while the filter relies on gravity and minutes of contact.

MethodDriving forceTypical contact timeTypical strengthHow it's drunk
Drip or pour-over (SCA Golden Cup)Gravity, controlled flow3–5 min1.15–1.35% TDSBlack, as is
South Indian filterGravity, slow drip through a packed bed10–20 minAbout 5–7% TDS (my estimate above)Cut 1:2 or 1:3 with hot milk
EspressoAbout 9 bar pump pressureAbout 25–30 sRoughly 8–12% TDSNeat or with milk

The espresso strength range is a common barista benchmark, not an SCA standard, so read it as a rough comparison. The filter's long contact also explains its flavour. A slow percolation at falling temperature through a dark, chicory-rich bed pulls a lot of roast bitterness and body and little bright acidity, which is exactly the profile that survives a big dose of milk and sugar. That is also why the dark roast matters; the roasting chemistry behind it is in the Maillard reaction in Indian cooking.

Questions people ask

How long should a South Indian filter take to drip?

Usually 10 to 20 minutes for a home filter. Much faster means the grind is too coarse or the bed channelled; much slower means the grind is too fine or the disc was pressed hard.

Why is my filter coffee decoction weak?

Most often the ratio: too much water for the powder. Weigh both once. Then check the grind and level the bed so water can't race down the sides.

Can I store decoction in the fridge?

Yes, for a day or so in a closed container, but aroma compounds fade and it tastes flatter. Warm the milk, not the decoction, to avoid boiling off more aroma.

Does chicory make filter coffee stronger?

It makes it darker and thicker, not more caffeinated. Chicory adds soluble carbohydrate and colour but no caffeine, so an 80:20 blend has about 20% less caffeine per spoon than pure coffee.

Is water temperature critical for filter coffee?

Less than grind and ratio. Sensory work on drip coffee found little difference between 87 and 93°C once strength and extraction were matched, so use freshly boiled water and focus on the dose.

More from the Taste Lab on Rasa.

Sources

  1. Batali, Ristenpart, Guinard: Brew temperature, at fixed brew strength and extraction, has little impact on the sensory profile of drip brew coffee, Scientific Reports 10:16450 (2020; source for Golden Cup ranges and temperature findings)
  2. Batali et al. (2020), free full text on PMC (source for Lockhart's 1957 Coffee Brewing Control Chart, its nine regions and the "ideal" centre)
  3. Wei et al.: Use of NMR-Based Metabolomics To Chemically Characterize the Roasting Process of Chicory Root, J. Agric. Food Chem. (2016)
  4. FSSAI: FAQ on coffee-chicory mixture standards, sub-regulation 2.10.4(1) (accessed 2026)
  5. FSSAI: Draft notice proposing 70% coffee in coffee-chicory mixtures (2018)
  6. Agro & Food Processing: FSSAI eases coffee-chicory labelling norms, extends deadline to July 2027 (2026)
  7. Coffee Board of India: Coffee statistics, production 2025-26 final and 2026-27 post-blossom, domestic consumption (2026)
  8. Coffee Board of India: Database on Coffee, July 2024 edition (2024)
  9. StoneX: India's Coffee Board issues first 2026 forecast (2026)
  10. StoneX: India's coffee exports set to exceed $2 billion in revenue (2025)
  11. Wikipedia: Indian filter coffee (construction, blends and serving) (accessed 2026)
filter coffeedecoctionextraction yieldTDSchicorySCA brewing control chartCoffee Board of Indiakaapi
Sapta

Sapta (Saptarshi Sarkar) is a mechanical engineer turned marketer in Bengaluru and a solo touring rider. He writes the seven Saptaverse blogs. Work with him or read more of Foodblog.

Keep reading in Foodblog

From the other blogs