Working on sustainability indicators in Latin American cocoa supply chains, one pattern becomes hard to ignore: the measurement burden accumulates downward.
A buyer adopts a standard. A cooperative adds the required data collection. Another buyer adopts a different standard with overlapping but non-identical requirements. The cooperative adds a second set of forms. Nobody removes the first.
The four dimensions, unevenly served
Frameworks like SAFA organize sustainability into four dimensions: good governance, environmental integrity, economic resilience and social well-being. In practice, coverage is lopsided. Environmental indicators are relatively well developed because they are easier to define and verify. Governance and social well-being tend to be represented by proxies that are cheap to collect rather than by measures that reflect the dimension.
This is not a failure of intent. It reflects what is measurable at reasonable cost — which means the framework quietly optimizes for what is easy rather than for what matters.
Three questions I use to sanity-check an indicator set
Who bears the collection cost, and were they consulted? If the producer or the cooperative technician carries the burden and had no say in the design, the data will be complete and unreliable — people fill in forms they do not believe in.
Is this indicator actionable at the level where it is collected? If the answer is only useful to someone three levels up, the person collecting it has no reason to be accurate.
Would removing this indicator change any decision? Most indicator sets could lose a third of their content with no loss of decision-making capacity. That third is pure cost.
Toward a lighter approach
The most useful sustainability assessments I have seen are not the most complete ones. They are the ones where each indicator has a named decision attached: this number, above this threshold, triggers this action.
Everything else is documentation. Documentation has value, but it should be budgeted honestly rather than smuggled into the producer’s workload under the label of measurement.