Theory of change: what it is really for, beyond the diagram
In most projects, the theory of change exists. It was produced during design, often in response to a donor requirement, validated alongside the project document, then filed away. Eighteen months later, few team members could find it, and fewer still would say it shapes any decision. What was delivered is a diagram. What was meant to be produced is a line of reasoning.
The confusion is understandable: the diagram is the visible part, the one that prints, presents well in a workshop and slips into an annex. But a chart of boxes and arrows has no value of its own. It is worth only what it forces people to spell out, and above all what it makes open to challenge.
What a theory of change makes explicit
Every project rests on a causal conviction. A health centre is built because proximity is believed to increase use of a service. Staff are trained because new skills are believed to change practice. Financial support is provided because capital is believed to be the binding constraint.
These convictions are rarely wrong. They are mostly just unstated. They circulate implicitly in the minds of the people who designed the intervention, and they stop circulating the moment those people move on.
A theory of change has one precise function: to put a project’s causal reasoning in writing, precisely enough that someone could contest it. Everything else follows. Once the reasoning is written down, it can be discussed with partners, tested against the field, revised when data contradict it, and handed over when the team changes.
Reasoning that cannot be contested cannot be improved either.
The six components
A complete theory of change articulates six elements. The first four are almost always present. The last two are what separate a line of reasoning from an organizational chart.
Component 04 deserves a word. A mechanism is not an activity, it is the reason an activity produces an effect. “Train the staff” is an activity. “Trained staff gain enough confidence to make a diagnosis they previously referred elsewhere” is a mechanism. The first is planned, the second is observed or disproved.
Theory of change and logical framework: two distinct questions
The question comes up in nearly every design workshop: should we do one, the other, or both? It is badly framed, because the two tools answer different questions.
That complementarity of course assumes the logical framework itself does not stay frozen after validation, which brings us back to its ability to evolve during implementation. Living reasoning plugged into a dead matrix produces nothing.
Assumptions, the only genuinely testable part
In a theory of change, only one thing can be confirmed or disproved by data: the assumptions. The problem is that they are almost always written in a way that makes this impossible.
“The political context remains favorable”, “partners remain engaged”, “beneficiaries buy into the approach”: these statements look like assumptions, but none of them points to a specific observation. No data will ever contradict them, so no data will ever validate them either.
This requirement is not theoretical. An assumption written this way points to the very data you need to track in order to know whether the project holds up. In the example above, two indicators fall out naturally: availability of test kits in facilities, and the screening refusal rate. Neither measures a project result, and yet they are the ones that will tell you first whether the reasoning is holding.
Using it during implementation, not only at design
A theory of change is treated as a start-up deliverable. That is precisely what makes it useless, because it is then produced at the moment the team knows least about the field.
A different use consists in bringing it back at three specific points in the life cycle.
At periodic reviews. The question is not only “where are we against targets?” but “which of our assumptions is the most fragile today?”. That question changes the nature of the meeting: it surfaces weak signals the dashboard does not show.
Facing an unexplained gap. When an indicator drops with no obvious reason, the usual reflex is to look for an implementation problem. The useful reflex is to go back to the reasoning: which step in the chain did not occur, and which condition was missing? This is exactly the logic behind choosing a method from the evaluation question.
Before a scale-up decision. Scaling up means betting that a mechanism observed in one context will repeat elsewhere. The theory of change is where that bet is spelled out, and where the conditions that will have to be met in the new territories are identified. That question connects directly with the level of evidence needed before a scale-up decision.
Five mistakes that make it useless
The fifth mistake is the costliest, because it cancels out the effect of the four previous fixes. A theory of change that is well built, participatory, explicit about its assumptions and reviewed every year is still worth nothing if no data ever puts it to the test.
Attach every assumption to tracked data
Delta Monitoring connects the assumptions of a theory of change to the indicators that put them to the test, and shows those indicators evolving through implementation. The reasoning stops being a design document: it becomes a set of assumptions that data confirm or contradict, while teams can still act on it.
A line of reasoning, not a deliverable
A theory of change is not one more document in the project file. It is the single piece that explains why the whole system makes sense.
A project whose reasoning is written down and tracked can be wrong, notice it, and correct course. A project whose reasoning stayed implicit can be wrong for three years without ever noticing, because nothing in its system is designed to tell it so.
The difference lies neither in budget nor in the sophistication of the methods. It lies in one simple discipline: having written down what you believed, and having planned a way to check whether you were right.
is proudly powered by WordPress
