Most quarterly donor reports we read still measure the same things they did a decade ago: number of boreholes drilled, workshops held, beneficiaries trained, agreements signed. These are outputs — discrete artefacts of activity that are easy to count and easy to claim.
What they tell you about whether the world is materially different at the end of the quarter is approximately nothing.
The trap of countable things
Outputs survive because they are convenient. A board meeting needs a deck; a deck needs numbers; numbers need to be available by the cut-off date. Outcomes — children who finished school, women whose income held up through the season, communities whose cholera rate stayed below baseline two rains later — are slow, fuzzy, attribution-fraught, and frequently embarrassing.
So the field has spent thirty years optimising the metrics it can produce on time, and reading like the metrics it produces are the work.
What we actually try to track
Across our projects we hold ourselves to a smaller set of questions:
- Did the system continue to operate after we left?
- Did the marginal cost of running it bend down or up?
- Did governance remain local, or quietly revert to the donor?
- Did adjacent systems (health, gender, livelihoods, climate adaptation) move with it, or did the project siphon attention away from them?
None of these reduce to a clean number on a deck. Each of them needs three years and several conversations to answer honestly.
What we ask of partners
If a funder needs us to optimise toward an output we don’t believe ladders up to a real outcome, we say so before the contract is signed. We have walked away from engagements over this. We’d rather be unfunded than complicit in another decade of well-counted boreholes that nobody is drinking from.
Our logic model is public for exactly this reason: it gives funders a chain of cause that they can interrogate. Inputs → activities → outputs → outcomes → impact. Outputs are a checkpoint. They are not the destination.
A small ask
The next time someone hands you a report, ask what changed at the outcomes layer. If the answer is "we’re still measuring outputs, we’ll have outcomes data in 2028", at least the conversation is honest.
That’s the bar.