Most evaluation projects begin with a spreadsheet. Ours begin with a conversation. Before we ask whether your programme works, we ask what you're trying to change, and for whom — and we keep asking until the answer surprises you, or us.
That's not a soft start. It's the part most teams skip, and it's where ninety percent of evaluation failure is born. Theory of change is a fancy phrase for a simple promise: that you can draw a line from what you do to what changes — and that we'll help you draw it honestly.
The honesty matters. A theory of change drawn after the fact, to please a funder, is a category error — it's a marketing diagram dressed up as a programme logic. We've all done it. We've all seen it.
The first meeting is the most important meeting
We block half a day for the kick-off. No agenda beyond a single question: what are you actually trying to change? Most teams get to the real answer in the second hour, not the first. The first hour is the version they've practised — the one with the funder's vocabulary in it.
The second hour is when the founder admits that drop-out is a symptom, not the problem; or that the policy brief was never going to land because the audience was wrong; or that the metric they've been reporting for three years isn't the metric anyone cares about.
A theory of change drawn after the fact, to please a funder, is a category error.
That's the meeting. Everything else — the design, the methods, the report — flows from whether you got the second hour right.
What a good theory of change does
It does three things, and only three:
- It names the change you're after, in language a stranger would understand.
- It draws the line from your activity to that change, with the assumptions on the line, not under it.
- It is short enough to put on one page.
We've seen theories of change that ran to twelve pages and contained four hundred boxes. They were diligent. They were also unusable. The diagram was so dense nobody on the implementing team could hold it in their head, which meant nobody could test it against reality, which meant it never adjusted when reality answered back.
Where the work goes wrong
Three places, in our experience.
The change isn't named clearly. "Improved outcomes" isn't a change. "Girls staying in school past fourteen" is. The first lets you off the hook. The second commits you.
The assumptions hide. Every arrow from activity to outcome carries an assumption — usually several. "Workshops change behaviour" assumes attendance, attention, retention, transfer, and a friendly environment for the new behaviour. Surfacing those assumptions is how a theory of change earns its keep; hiding them is how it becomes decoration.
The diagram doesn't change. A theory of change is a hypothesis, not an artefact. If you haven't revised it after a year of running the programme, either the world is unusually compliant or you've stopped paying attention.
The honest part
When we say we listen first, we mean it as a discipline, not a slogan. We don't run a workshop on day one because we don't yet know enough to facilitate one. We sit in your office. We read what you've written. We ask the team what they'd do if the funder disappeared tomorrow.
Then — and only then — we draw the diagram. And we draw it lightly enough that it survives contact with what we learn next.
Field note №04 · An earlier version of this essay appeared as a memo to a client in 2024. Republished with permission.