A theory of change (ToC) is an explicit, testable account of how and why a set of activities is expected to lead to long-term change. It connects what you do to the change you seek — and names the assumptions that have to hold for that chain to work.

The ToC approach grew out of programme evaluation. Carol Weiss argued in 1995 that complex community initiatives kept failing evaluation because their underlying theories — the assumptions about how and why they would work — were left unstated and untested.

A ToC is always anchored in a specific context — the place, the people, the politics, the existing services. The same activity can produce different outcomes in different settings, so a ToC that travels well names the context it assumes.

Every programme already has a theory of change — in someone's head. The discipline of a ToC is to drag it into the open where the whole team can see it, argue with it, and improve it before the money is spent.

Programme staff, finance, M&E, partners and the community often hold different, unspoken theories of how the work will help. A shared ToC surfaces those differences early — while they are cheap to resolve.

Laying out the pathway makes missing steps and leaps of logic visible. If training teachers is supposed to improve learning, the ToC forces you to show every link in between — and to notice where one is missing.

A ToC is the backbone of theory-based evaluation. Each outcome on the map is something to measure; each assumption is something to test. Evaluation stops being a verdict and becomes an inquiry into which links held and which broke.

A ToC clarifies your sphere of control (activities and outputs), your sphere of influence (outcomes you contribute to), and the sphere of interest (impact shaped by many actors). It keeps you from over-claiming credit.

ToC, the logframe and the results chain are cousins, not rivals. They share the activities–to–impact logic but differ in form, richness and what they make visible.

A filled-in logframe is not a theory of change. The matrix can be complete while the causal reasoning — the ‘because’ arrows and the assumptions that connect the rows — remains entirely unstated.

Outcome Mapping (IDRC) and Outcome Harvesting are kindred approaches that focus on changes in the behaviour of actors a programme influences. They pair well with a ToC, especially for advocacy and systems change where impact is hard to attribute.

Impact is the long-term, often population-level change a programme contributes to — reduced child stunting, an educated generation, safer cities. It usually unfolds after the project ends and is shaped by many actors beyond you.

The signature technique of ToC, from ActKnowledge and the Aspen Roundtable, is backwards mapping: begin with the long-term change you want, then work backwards asking ‘what must be true just before this, for it to happen?’

Backwards mapping produces a stack of preconditions. Read top to bottom it is a wish; read bottom to top it becomes a causal pathway.

Real pathways branch. Reaching grade-level reading may need three parallel pathways — better teaching, regular attendance, and a literate home environment — that converge on the goal.

Keep mapping backwards until you reach outcomes your programme can plausibly act on with its activities. Below that line are your entry points; above it are the changes you are trying to set off.

At each step, ask two questions: are these preconditions necessary (must they all be true?) and are they sufficient (if all are true, does the next outcome follow?).

Every arrow in a ToC hides an assumption: a belief about why one outcome will lead to the next. Make those beliefs explicit and the ToC becomes testable; leave them buried and it becomes a leap of faith.

A ToC arrow is stronger when backed by a rationale — evidence, theory or experience that the link tends to hold. Cite research, prior evaluations, or the lived knowledge of the community.

A failed assumption is a risk by another name. Listing assumptions doubles as a risk register: plot each by how likely it is to fail and how badly the pathway breaks if it does.

Some assumptions are load-bearing: if they fail, the whole theory collapses. A women's-livelihood ToC may rest on the assumption that households will let women keep their earnings — the quiet hinge on which everything turns.

An explicit assumption is a question you can answer with evidence. Build a small check for the riskiest ones into your monitoring, so you learn early whether the theory is holding — not at the final evaluation.

Every outcome on the map needs an answer to one question: how will we know this change occurred? The answer is an indicator — an observable marker that the outcome is real.

An indicator is only meaningful against a baseline (where you started) and a target (where you aim to reach). Without a baseline you cannot show change; without a target you cannot judge whether it was enough.

For each indicator, name the means of verification — the data source and method. An indicator with no realistic source is a wish, not a measure.

It is tempting to attach three indicators to every box. Resist. Each indicator costs time, money and respondent goodwill to collect. Choose the few that genuinely move decisions.

The visual heart of a ToC is the outcomes map: each box is an outcome (a change), each arrow a causal claim, read upward from entry-point activities to the long-term goal.

The diagram needs a narrative that walks through each pathway: what changes, why we believe it will, what evidence supports the link, and what assumptions must hold. The narrative is where nuance lives.

A ToC drawn alone is a document; a ToC drawn with the team, partners and community is a shared belief. The workshop — arguing, re-ordering, disagreeing — is where the real value is created.

Communities hold knowledge about what actually drives change in their lives — knowledge no logframe captures. Involving them improves the theory and grounds it in lived reality, not assumption.

A ToC is never finished. Draft it quickly, use it, and revise as you learn. An over-polished first version signals false confidence; a rough, living one invites the questioning that makes it strong.

A map with sixty boxes and a hundred arrows communicates nothing. The discipline is to show the main pathways clearly and push the detail into the narrative and the indicator table.

Activity-itis is the most common disease of ToCs: the map fills with things the programme will do — train, build, convene — and never reaches the changes those activities are meant to cause.

An arrow with no stated assumption is a claim nobody has examined. The link ‘women earn income → women are empowered’ hides a huge assumption: that they control how the income is spent.

The opposite failure: a map so dense with boxes and crossing arrows that no one can follow a single pathway. Complexity is not rigour — often it hides muddled thinking behind a busy picture.

At design, the ToC stress-tests the programme before money is spent: are the pathways plausible? Are activities matched to the change sought? Where are the riskiest assumptions, and can the design strengthen them?

In MEL, the ToC turns into a measurement plan: each outcome becomes an indicator, each assumption a monitoring question, each pathway an evaluation hypothesis to confirm or revise.

Where you cannot run a controlled experiment, contribution analysis uses the ToC as its spine: gather evidence on each link, check whether the pathway played out as theorised, and rule out alternative explanations.

When monitoring shows an assumption failing, the response is not to bury it — it is to adapt. The ToC becomes the place where learning is recorded and the programme's direction is consciously adjusted.

A ToC structures reporting around change, not activity. Instead of ‘we trained 500 people’, you can say ‘here is the change we contributed to, here is the evidence, and here is where the theory did and did not hold.’

Imagine a district programme aiming to reduce low birth weight by improving maternal nutrition. Let us build its theory of change end to end — goal, pathway, assumptions and indicators.