Setting Indicators That Are Still Usable Two Years Later

Ask most program teams two years into delivery whether they are still tracking the indicators listed in their original proposal, and the honest answer is usually partial. Some were dropped when the data source turned out not to exist. Some were replaced informally without anyone updating the results framework. A few are technically still being tracked, but by a different method than what was specified, which quietly breaks any baseline comparison. None of this is unusual. It is what happens when indicators are chosen to look complete on a template rather than chosen to be collectible.

Why indicators go stale

Three habits produce this outcome reliably. The first is selecting an indicator because it matches a funder's standard results framework, without confirming the program actually has, or can build, a data source for it. The second is skipping the baseline. An indicator with no baseline value cannot demonstrate change, only a single point in time, which means the endline number arrives with nothing to compare it against. The third is changing an indicator mid-program to reflect what turned out to be easier to measure, without documenting the change, so a later evaluator cannot tell whether a shift in the numbers reflects a real change in the program or a change in what was being counted.

What makes an indicator durable

A durable indicator has three properties. It is tied to a data source that genuinely exists or can be built at a defensible cost, not one assumed to exist because a similar program elsewhere used it. It is feasible to collect on the program's actual reporting cadence, whether that is monthly, quarterly, or tied to specific programmatic milestones, rather than a cadence borrowed from the funder's own reporting calendar. And it is agreed with the people who will actually be responsible for collecting it before the program starts, not handed to them afterward as an obligation someone else designed.

Performance Measurement Framework development exists specifically to force this conversation early: what will be measured, how, by whom, and how often, confirmed with the field team before the indicator is locked into a funder agreement.

Close-out is not when to start counting

The deeper fix is sequencing. When indicators are agreed at the design stage and data collection begins alongside program delivery, evaluation becomes a matter of pulling together information that was already being gathered. When indicators are treated as a compliance section to complete before disbursement and then left alone, evaluation becomes a reconstruction exercise, built from whatever partial records survived two years of program turnover. The second version is more expensive, less reliable, and arrives too late to inform anything except the final report.