Radhika Agrawal
Impact tool

Theory of Change Stress-Test

Everyone can fill in the boxes — problem, activities, outputs, outcomes, impact. A theory of change doesn't break in the boxes; it breaks in the arrows between them. This tool red-teams those causal links: it finds the missing middle, the unstated assumptions, the attribution overreach — and ranks the assumptions most likely to be challenged.

Built by Radhika Agrawal · a critic for a theory of change you already have, grounded in contribution analysis. It stress-tests logic; it is not an evaluation.

How to use it. Bring a theory of change you already have. Drop each level into the chain below — the example is pre-filled so you can see the shape — then work down the three links between them. Nothing is stored or sent; it all runs in your browser.

The yardstick. Every arrow hides an assumption that must hold. Following John Mayne's contribution analysis, an assumption is only as strong as what backs it: general agreement, strong logic, action being taken to secure it, or prior evidence. Logic on its own is the weakest footing; evidence and active effort are the strongest. The tool scores each link on that basis and shows you which arrow will be challenged first.

Your results chain

Keep each line to a sentence. The three arrows between the last four levels are what gets tested.

Problemthe start
Activitieswhat you do
Outputsimmediate
Outcomesthe change
Impactlong term
Stress-test the three links

For each arrow: name the assumption it rests on, tick what actually backs that assumption, and give one honest gut-check.

Where it will be challenged

Your chain, by link strength

The method
Boxes vs. arrows. A theory of change is a chain of causal claims. The boxes (what you do, what you hope) are easy; the arrows — the leaps between them — are where the logic is load-bearing and usually untested.
The missing middle. The output→outcome jump is where most theories of change fail: "we ran the workshops" is an output, "behaviour changed" is an outcome, and the arrow between them is often hope, not mechanism. This link is weighted heaviest here.
Contribution, not attribution. Most programmes are one contributory cause among many — neither necessary nor sufficient on their own. Claiming an impact outright, where many forces are at work, is the most common overreach (John Mayne, contribution analysis).
Rival explanations. A robust theory of change names what else could have produced the result, and rules those out where it can. If nothing could ever disprove your chain, it isn't a theory — it's a wish.
The four grounds. An assumption is plausible only through agreement, logic, action or evidence. A link resting on logic alone is fragile; add evidence and active effort and it firms up.