Why and how to disaggregate an indicator

Disaggregation means breaking an indicator's total down by a meaningful category, rather than reporting only the aggregate. A programme can report "78% achievement" on an indicator overall while badly underperforming for women, for one district, or for one age group — and the aggregate figure alone would never reveal that. Disaggregation exists specifically to surface exactly that kind of hidden imbalance.

The most common disaggregation dimensions in programme work are sex/gender, age group, and geography (region, district, or lower). Where a programme delivers through more than one channel, disaggregating by Implementing Partner or by service provider shows whether performance is even across delivery channels or concentrated in one. Many programmes also need a dimension specific to their own content — a value chain, a crop type, a service type — that only makes sense for that particular sector.

A disaggregated breakdown is only trustworthy if it reconciles: the sum of a "by gender" breakdown, for instance, has to equal the same total the headline indicator itself reports for that period. A disaggregation that is calculated separately from the headline figure, by a different process, can silently drift out of agreement with it — which is a strong (and common) signal that something in the underlying data pipeline is broken.

A related and frequently confused distinction is unique people reached versus service or event counts. If a beneficiary attends three training sessions, that is three service events but one unique person reached. Reporting service events as if they were unique people overstates reach — sometimes dramatically, on a programme with high repeat engagement — and is one of the most common inflation points in delivery reporting.

Disaggregation groups the same underlying records by a categorical field already present on the data source — a form field, a beneficiary attribute — rather than requiring a duplicate, hand-maintained breakdown.

Grouped totals are computed with the same SUM/COUNT DISTINCT logic as the indicator's own headline figure, so a "by gender" breakdown always reconciles back to the same total the indicator itself reports — and reach figures always distinguish unique beneficiaries from service/event counts rather than conflating them.