Skip to content
ViableDevelopment Associates
Article · Monitoring

Why logframes break at the outcome level — and how to fix them

Viable Development Associates Limited6 min read

Most results frameworks are strong on activities and outputs and weak exactly where it matters. Three habits that close the gap.

Open almost any programme logframe and you will find the same pattern. The activity and output rows are crisp: number of people trained, number of facilities equipped, number of meetings held. Then the outcome row arrives and the language softens — 'improved capacity', 'strengthened systems', 'increased awareness'.

This is not a drafting problem. It reflects a genuine difficulty: outcomes are changes in behaviour, practice or condition that the programme influences but does not control. They are harder to define, harder to measure and slower to appear. But a framework that cannot measure outcomes cannot tell a programme whether it is working.

1. Name the behaviour, not the capacity

'Improved capacity of health workers' cannot be observed. 'Health workers correctly classify and treat childhood pneumonia' can. Every outcome statement should describe something a person or institution does differently — something you could, in principle, watch happen.

2. Write the assumptions down — and monitor them

The jump from output to outcome always rests on assumptions: that trained staff stay in post, that equipment is maintained, that communities trust the service. These are usually listed once in an annex and never revisited. We recommend picking the two or three riskiest and tracking them as lightly as possible — a short question in a routine survey is often enough.

3. Accept proxy measures, deliberately

Perfect outcome measurement is expensive. A well-chosen proxy, collected consistently, is worth more than an ideal indicator that is measured once at endline. The key is to document why the proxy is reasonable and what it might miss.

A framework that cannot measure outcomes cannot tell a programme whether it is working.

None of this requires a new framework. It requires spending a little more time on the middle of the results chain — where most of the real uncertainty lives.

Keep reading

Let’s design the evidence you need.