Published production where it exists — and the gaps where it does not.

There is no modelled production. Nothing here is allocated from a national total, interpolated between neighbours, or derived from a yield assumption. If an institution published a figure for a region, it is here with its citation. If nobody did, the region is blank.

There is no confidence score. A number out of a hundred with no stated derivation is precision theatre. What you get instead is the source, the publication date, and whether it measures production or purchases — which is what actually determines how much weight to give it.

There is no yield unless one source published both sides of it. Yield computed from a production figure and an area figure taken from different documents is arithmetic on unrelated numbers. Nigeria has real yields because the EU value-chain mapping gives production and hectares together; Ghana does not, so Ghana shows none.

Why the country shading uses round numbers

The bands are 10, 50, 250 and 1,000 thousand tonnes rather than equal slices of the data. Cocoa production is extraordinarily concentrated — Côte d'Ivoire grows roughly fifty times what Uganda does — and equal-count bands flatten exactly that, putting the largest producer on earth and a marginal one in near-identical tones while making a thousand-tonne grower look consequential. Only Côte d'Ivoire sits in the top band. That is the correct picture.

Three layers, three things they mean

The country layer comes from FAOSTAT: every producing nation, back to 1961, free and redistributable under CC BY 4.0. It is the only comparably-defined global series that exists. Two things to hold in mind when reading it. FAO reports on calendar years, so a FAO 2024 figure is not the 2023/24 crop year. And FAO's West African numbers are largely national-agency submissions, so for Ghana and Côte d'Ivoire they trace back to COCOBOD and the CCC — they are not an independent check on those bodies.

The regional layer is what sits on top: sub-national figures where an institution has published them. That is where the interesting information is, and it is thin, which is the honest state of public knowledge about where cocoa actually comes from.

The department points are a third thing entirely, and drawn as circles rather than shaded areas for a reason: Trase reports by department while our boundaries are regions, and painting a department's volume across its whole region would claim far more ground than the figure covers.

Ground truth, and why it is drawn differently

The dashed green outlines mark Ghanaian regions where surveyors have physically confirmed cocoa. Western Ghana carries 7,142 verified observations; Central carries 2,487. No marketing board and no model is involved — someone went there.

It is drawn as an outline rather than a shade on purpose. These polygons are a training sample collected to teach satellite classifiers, not a census, so the hectares are what surveyors mapped and not the region's cocoa area. Shading by them would read as an area estimate, which is precisely the misreading the dataset itself warns against. What they do support is confidence: a region with seven thousand confirmed observations bears a stronger claim than one with forty.

Comparing seasons

The comparison shows only regions carrying a figure in both seasons, and states how many do. Change is always computed from the earlier season to the later one regardless of which you selected — an early version reported Ghana's Ashanti decline as a 67% rise because it ran the arithmetic backwards, which is worse than showing nothing.

One caution the map repeats where relevant: Ghana's regional boundaries changed in 2018, so a unit with the same name in two seasons may not be the same geography.

What the gaps mean

Côte d'Ivoire is nearly empty, and the reason is the finding. It grows about 36% of the world's cocoa, and no institution publishes departmental production. What does exist is traceability: Trase reports that only 48% of 2024 export volume could be traced back to a specific department of production from what trading companies disclose. Thirty-four percent moved through intermediaries and 18% was simply not disclosed.

So the three green circles on the map are not production. They are the volume of EU-destined cocoa traceable to that department — San-Pédro 52,700 t, Gagnoa 36,700 t, Duékoué 33,600 t in 2024. A department could produce far more than its circle and much of it would be invisible, because nobody disclosed where it came from. Half of the EU's directly-traceable supply comes from eight departments.

Trase adds a caution worth repeating: the traceable share has risen partly because their methods for locating cooperatives improved, not because the sector got more transparent. That distinction is the difference between a real change and a measurement artefact, and most reporting collapses it.

The practical consequence for anyone with a December 2026 EUDR deadline: roughly half of Ivorian export volume currently has no departmental origin on record, let alone the plot-level geolocation the regulation asks for.

Ghana's boundaries changed underneath the data. The Cocoa Board reported by ten regions until 2018 and by sixteen since, splitting Western into Western North and Western South. A 2017 figure for "Western" and a 2023/24 figure for "Western South" are not the same geography, so this map keeps them in separate seasons and does not restate one into the other.

Purchases are not production. The 2019/20 rows are COCOBOD purchases — beans bought through licensed buying companies. Purchases fall when buyers run short of cash even when the crop is unchanged, which is precisely the gap our field panel exists to measure. The map labels which is which and never adds them together.