Coffee went into surplus for the first time in five years. It is worth half a cent a cup to you

Short answer, because it is the question everyone in the trade is being asked right now: the world coffee price averaged 287.29 US cents a pound in August 2026, which is 3.3% below where it was a year earlier. That is the whole dividend from a record Brazilian crop and the first global surplus in five years, and in a double shot it comes to about half a cent. The number worth your attention sits three pages further into the same report, and it is not a price at all.

Where the price actually is

The ICO Composite Indicator Price, the I-CIP, is the industry's benchmark for green coffee. It is published monthly and quoted in US cents per pound. Here is what it has done, taken straight from the ICO's own monthly reports:

• August 2025: 297.05.
• June 2026: 248.90, and on 9 June it touched 231.96, the lowest level in nearly two years.
• July 2026: 287.26. A jump of more than 15% in a single month.
• August 2026: 287.29. Effectively flat, ranging between 279.38 and 301.97 during the month.

So the fall happened, and then most of it was handed back inside four weeks. Anyone who told you in June that relief was on its way was right for about three weeks.

The reason it has not stayed down is stock timing rather than crop size. The ICO notes that certified Arabica stocks are historically low, and that futures stayed in backwardation through August, meaning coffee for delivery now costs more than coffee for delivery later. That is the market saying there is plenty of coffee coming and not much of it is here yet. In June, US certified Arabica stocks fell 13.3% to 0.41 million bags, the lowest since February 2024.

A surplus on a spreadsheet and a full warehouse near your roaster are two different things, and it is the second one that sets your invoice.

One kilo of green coffee, in cups

This is the bit that usually goes missing, so let us do it on real numbers.

At 287.29 US cents a pound, a kilo of green coffee costs $6.33. Roasting burns off water, so assume a roast loss of about 18% and a dose of 18 grams for a double shot. Both are rules of thumb rather than your numbers, but they are in the normal range.

One kilo of green coffee therefore makes about 45 double shots, and the green coffee inside each one costs about 14 US cents.

Now run the same sum across the whole year:

• At last August's price, 297.05, it was 14.4 cents.
• At August 2026's price, 287.29, it is 13.9 cents.
• At the two-year low of 231.96 on 9 June, it was 11.2 cents.

The entire twelve-month move in the world coffee price is worth about half a cent per double shot. The full distance from the two-year low to today is worth under three cents. The ratio holds whatever currency you bill in, because both sides of it move together.

That does not mean your bean bill is irrelevant. Across 60,000 coffees a year, three cents is $1,800, which is real money. It means the bean is not the thing deciding whether this year works.

The line in the same report nobody quoted

Here is what was published on 10 September alongside the surplus.

World coffee production is estimated to rise 4.4% to 183.6 million bags in coffee year 2025/26, after rising 2.5% to 175.9 million the year before. Fine, that is the supply story and it has been well covered.

Consumption is the other half. It rose 4.3% to 182.2 million bags in 2024/25. For 2025/26 the ICO estimates it will fall 0.8%, and it names where: a reduction in the United States, the world's largest coffee-consuming country.

That is the first thing in this whole release that should change a decision. The surplus is not only a bumper harvest. It is partly people buying less coffee after two years of paying more for it.

The USDA's own coffee circular, published in July, points the same way for the year just ending. It revised 2025/26 imports down by 700,000 bags for the United States and 500,000 bags for the European Union, and in both cases gave the same reason: lower consumption.

In fairness, the two bodies do not agree about what happens next. The USDA forecasts global consumption reaching a record 179.7 million bags in 2026/27, with the largest gains in the EU and the US, on the back of world production hitting a record 189.7 million bags. They use different definitions and different coffee years, so the two sets of numbers are not directly comparable. The honest summary is that both agree demand softened in the year now ending, and they disagree on whether it comes back.

If it is not the bean, what is it

Take one staffed hour behind the machine. Say that person costs you $18 an hour all in.

• Busy hour, 25 coffees served: the labour in each cup is 72 cents.
• Quiet hour, 15 coffees served: the labour in each cup is $1.20.

Same barista, same wage, same coffee. The cost per cup moved 48 cents because the queue moved. That is roughly a hundred times the effect of a year's worth of world coffee price movement, and it is entirely inside your control through opening hours, rostering and how many people you have on at 3pm on a Tuesday.

Rent behaves the same way. It does not care how many cups you sell, so every cup you do not sell makes the rest more expensive.

Do it on your own paper this week. Take last week's total wage cost, divide it by the number of drinks you sold, and write that number next to 14 cents. Then do the same with a week's rent. Most owners have never seen those three numbers side by side, and it reorders the to-do list permanently. The break-even calculator will do the heavy version, and staff scheduling and labour cost covers how to move the rostering half.

Two questions before your next roaster contract

You do not buy green coffee, you buy roasted, and the two are not on the same clock. Your roaster bought their beans months ago and is carrying a contract of their own. So a monthly world price is not a stick to beat them with, but it is a reason to ask better questions.

One: what is my price actually made of? Ask them to split it into the green cost and everything else, meaning roasting, packing, delivery, equipment and service. If a rise came through in the last eighteen months and the green half has since fallen, that is a fair conversation to have. If they will not split it, that tells you something too.

Two: what am I tied to, and until when? Many cafes are on an equipment loan dressed up as a bean contract, where a machine was supplied free against a minimum weekly kilo commitment. That is not necessarily a bad deal, but you should know the remaining term, the minimum volume, and what it would cost to leave. If demand does soften, a minimum volume commitment is the clause that hurts, and it is easier to renegotiate when you are buying well than when you are behind.

What two years of gross margin says about you

If you ever sell, a buyer will look at your gross margin across 2025 and 2026 together, and those two years happen to be an unusually good test.

One cafe held its margin through a genuine cost shock by lifting its average ticket and its throughput. Another let the margin sag in 2025 and got some of it back in 2026 when the world price dipped. Same margin today, completely different businesses. The first one proved it has pricing power. The second one proved it has a commodity exposure and got lucky.

Diligence separates those two in about an hour, and it pays for the first one. So the more useful question is not whether beans get cheaper. It is whether you can show, in your own numbers, that your margin survived the worst coffee market in a decade. If you cannot answer that yet, improving profit margins is the place to start, and the pricing power calculator will tell you which items on your board can carry a rise without costing you volume.