Theory of change: a practical guide with worked examples

A theory of change is a written account of how an activity is expected to produce a result. It sets out what you put in, what you do with it, what that produces, what changes for the people involved, and the longer-term difference that contributes to — and, crucially, why each step should lead to the next.

That last part is what separates it from a list of boxes. Anyone can draw a chain of arrows. A theory of change states the reasoning behind each arrow, so that a reader can disagree with it. If nobody could possibly disagree with your theory of change, it is probably describing activity rather than causation.

Most UK funders now expect one, either by name or in substance. The National Lottery Community Fund asks how you know your approach will work. NIHR asks for a logic model and an evaluation plan. Trusts and foundations ask what difference the money will make and how you will know.

The reason is that funders have learned to distinguish between organisations that can describe their activity and organisations that can explain their effect. The first group reports how many sessions it ran. The second reports what changed, and can say how confident it is.

A theory of change is also the document that decides your evaluation. The outcomes you write into it are the things you will later have to measure — which is why it is worth writing with the measurement in mind rather than as a bid-writing exercise.

Nearly every version uses the same five levels. The names vary; the logic does not.

The most consequential line in the whole diagram sits between outputs and outcomes, because that is where you stop describing what you did and start claiming what you caused.

Note the language shifts with control. Outputs are things you deliver. Outcomes are things you contribute to. Claiming an impact you caused single-handedly is the fastest way to lose a sophisticated reader.

These are used interchangeably and should not be. The distinction is simple and it matters when a funder asks for one specifically.

In practice a theory of change contains a logic model. If a funder asks for a logic model, the chain alone is enough. If they ask for a theory of change and you send a chain with no reasoning, you have answered a different question.

The common mistake is to start with activities, because activities are what you know. That produces a document which justifies what you already do. Start at the other end.

Between every pair of levels sits an assumption — something that must be true for the arrow to hold. Writing them down is what makes the theory testable, and it is the first thing an evaluator looks for.

Take a mentoring programme claiming that mentoring hours lead to improved attendance. The arrow carries at least four assumptions: that the young people most at risk will actually attend; that mentors are matched well enough to build trust; that attendance is affected by confidence rather than by circumstances outside school; and that three terms is long enough for a change to show.

An outcome that cannot be measured cannot be evidenced, and a theory of change full of them produces an evaluation that reports activity instead of effect. Each outcome needs three things settled before the work starts.

There is also a question of numbers. With thirty participants you can describe what happened; you cannot reliably detect a small change, and reporting one as though you had is the error most often picked up by a critical reader. Working out in advance how many you need is a sample size and power question, and it is far cheaper to answer at bid stage than after the grant is spent.

The same structure, in three sectors. Each is deliberately short — a real one would carry assumptions and indicators alongside.

Notice the riskiest assumption row. In each case it sits outside the organisation's control, and in each case it is the thing most likely to decide whether the programme works. That is normally true, and it is why the assumptions are worth writing.