A tea broker in India earns, by the reckoning of the credit-rating agency CRISIL, a commission of around 1 percent on the value of every lot it sells at auction. In Colombo, the arrangement is different in its mechanics but similar in its size: a broker there keeps 1 percent of the 10 percent deposit a buyer must post within a week of winning a lot. Neither figure has moved much as the sale itself has gone from a room full of shouting men to a screen. That is the puzzle worth sitting with. Trading floors across finance and commodities have spent thirty years stripping out intermediaries the moment a computer could do the matching instead. Tea has had every chance to do the same, and largely has not.

The short answer is that the commission was never really the fee for running the sale. It is the fee for a bundle of judgment and custody that a matching engine does not supply: valuing a lot honestly, keeping the industry's word once a bid is accepted, and financing the gap between harvest and payment. How the auction itself works, garden to gavel, is the subject of its own reference; what follows is the narrower question of what the broker's cut is actually paying for, and why two of the world's oldest exchanges have kept paying it even after the technology to route around it existed.

What the commission is actually buying

A broker's job starts long before a buyer ever sees a lot. Producers hand over made tea, and the broker catalogues it by garden, invoice, and grade, draws a sample from every consignment, and tastes it against the standard for that grade before a single buyer is invited to bid. That tasting is not decoration. It sets the reserve price and the description the whole trade will act on, and a broker who grades a lot wrong is the one whose name is attached to the mistake.

After the sale, the same broker's name is on the paperwork that follows the tea out the door: the invoice, the account sale sent back to the producer, and, when a buyer later disputes the quality or quantity delivered, the claim that has to be settled. CRISIL's own rating report on J Thomas and Company, the oldest and by its reckoning still the largest tea broking house in India, lists a further, less obvious revenue line alongside brokerage and selling charges: interest earned on advances. A broker with a long relationship to a garden will extend it money against tea not yet sold, which means the commission is also, in part, the price of a credit line no bank extends on the same terms, because no bank tastes the tea.

None of this shows up on the sale-day ticker. It shows up in the fact that a grower in Assam and a blender in Hamburg, who have never met and have no reason to trust one another, both accept a number that a licensed third party put on the tea between them.

The figures, side by side

India and Sri Lanka charge for the same bundle of work in different currencies of measurement, which makes the comparison useful rather than just a pair of similar-looking percentages. J Thomas alone auctioned tea worth 3,663 crore rupees (roughly US$490 million at that year's exchange rate) in the 2022 financial year, CRISIL's rationale records, a bit over a third of all tea sold at Indian auction that year. On roughly 1 percent commission, that one firm's brokerage income alone runs into tens of millions of dollars a year, spread across six auction centres and a firm CRISIL traces back some 160 years.

Colombo prices the same function through the buyer's deposit rather than the seller's sale value: a winning bidder posts 10 percent of the lot's price within a week, and the broker keeps 1 percent of that deposit as its fee. It is a smaller absolute cut per lot than the Indian model, spread across the roughly 300,000 metric tonnes that pass through the auction each year, sold by eight competing broking houses rather than one dominant one. Different arithmetic, same underlying trade: a small, standardised cut of a very large, very repetitive flow of transactions, in exchange for grading, custody, and a guarantee neither party could credibly make alone.

Screens took over the sale, not the broker

India moved its national auction onto an electronic platform years ago, and Colombo, which had run its sale by open outcry since 1883, forced its own hand in April 2020, when a coronavirus curfew closed Sri Lanka's Chamber of Commerce and the industry needed a way to keep selling an essential crop. Eight brokers and roughly 300 buyers logged into the first digital sale that month, and the association's chairman, Jayantha Karunaratne, called it lower cost, more efficient, and more transparent, while conceding that changing the habits of some of the room's oldest hands would not be easy.

What did not change was who did the grading, the cataloguing, and the vouching. A screen can display a catalogue and take a bid faster than a room can shout one, but it cannot taste a sample, and it cannot put its name behind a claim when a buyer says the tea that arrived is not the tea that was described. That is the specific work the commission has always paid for, and digitising the bidding step left it entirely intact.

India is not finished changing its own screen, either. As of this writing, the Tea Board has a Kolkata firm building a third-generation, cloud-hosted auction platform to replace the current app, under a build-own-operate contract, reports the Hindi-language agricultural outlet Kisan India. The trade's reaction, as that outlet records it, is not enthusiasm. Anil George, chairman of Kochi's Tea Trade Association, is quoted calling the existing system "reliable and effective," and other voices warn that a changeover risks the kind of technical glitch that stalls a live sale. Nobody quoted is asking to remove the broker from the process. The argument, as in Colombo, is only ever about which version of the screen the broker should be working from.

Why one exchange brought the room back on purpose

The clearest evidence that the broker's value was never really about the bidding mechanism came on 31 January 2025, when the Colombo Tea Traders' Association held a live outcry sale for the first time in four years, 948 lots and nearly a million kilograms of tea, back in the same Chamber of Commerce room the pandemic had emptied. The association committed to holding one such live sale every quarter going forward, alongside the digital platform that still handles the bulk of the trade the rest of the year.

Nobody claimed the screen had failed. The stated reason was the opposite: to keep the relationship among brokers, buyers, and sellers alive in a form email and a bidding interface do not replicate, in an exchange old enough that its own association counts its history past 135 years. An industry given a genuinely faster, cheaper way to run its sale, one it had already adopted under real pressure, chose to spend one afternoon in thirteen putting people back in a room together anyway. That is a business voluntarily re-adding the cost of the thing it had just proved it could do without, because the thing it was buying back was never the auction mechanism. It was the room.

The rule that keeps half the crop flowing through a broker regardless

Not every part of the answer is about what a broker earns its keep by doing. Some of it is regulatory, and India is the clearest case. Since a 2015 amendment to the Tea (Marketing) Control Order, manufacturers there have been required to sell not less than 50 percent of their annual output through a licensed public auction, down from a 70 percent threshold set when the rule was first written. In March 2026, the Tea Board reissued the circular to every manufacturer in the country after a compliance review of 2024 and part of 2025 found widespread non-compliance, some of it from manufacturers who told the Board they had simply not known the rule existed.

The mandate exists because a public auction produces a transparent, published price that a wholly private sale does not, and the Board treats that transparency as worth protecting even when it is inconvenient for the seller. It is also, not incidentally, a rule that guarantees brokers a floor of business regardless of where the market would otherwise send the tea. Producers and brokers have disagreed about exactly this for years: brokers argue for routing more of the crop through the open auction, on the grounds that it improves price discovery and competition, while large estates have pushed the other way, reporting that private sales realise 7 to 10 rupees more per kilo (roughly 8 to 12 US cents) than the auction floor. Both sides are describing the same fact from opposite ends: the auction, and the broker who runs it, exist in their current form partly because the regulator has decided the transparency is worth more than the extra rupees a grower could otherwise keep.

Who the arrangement actually serves

Put the two threads together and the persistence stops looking like inertia. The commission survives the move to a screen because a bid interface never did the part of the job that justified the fee: the tasting, the cataloguing, the custody, the credit, and the name attached to a claim. And in India, at least half the crop keeps moving through that structure whether or not it is the seller's first choice, because a regulator has judged the public price more valuable than the private one. Neither of those is a story about nostalgia, whatever Colombo's quarterly outcry sale might suggest at first glance. It is a story about which parts of a trade a computer was actually asked to replace, and which parts nobody asked it to.