A standard theory of change stops at the causal chain and an end-of-programme evaluation. We add two sharp steps: designing the behaviour itself, and thinking through what the intervention causes elsewhere before it launches. Skip one of the four steps, and the theory stays a diagram that doesn't survive contact with reality.
1
Diagnose: which behaviour needs to change?
Start with what the target group actually does, not with the assumption written into the logic model. Which behaviour needs to shift, and which capacity, norms and friction hold the current behaviour in place?
2
Design: make the intended behaviour the easiest choice.
You design the defaults, the enrolment process and the social norm so the intended behaviour costs less effort than the current one, rather than relying on the stated incentive to do the work.
3
Predict: trace it through to the second and third order.
This is the step most theory of change work skips. What new behaviour does your intervention provoke elsewhere? Does the target group shift to a different channel, a different provider, a different part of the system? Map it in advance and the funder's next question won't catch you off guard.
4
Evaluate: measure behaviour, not activity delivered.
This is where evaluations often fall short. Checking whether the activities were delivered as planned measures nothing about the actual outcome. Pick a behavioural indicator upfront, set a baseline, and check after a few months whether behaviour actually shifted.
