Why Good Plans Fail Without a Theory of How Change Actually Happens

The uncomfortable truth about improvement

We love outcomes. We say we want less poverty, better health, safer neighborhoods, stronger schools. Then we build a program, fund a project, launch an initiative, and wait for the numbers to move. When they do not, the default explanation is usually some version of: not enough resources, not enough time, not enough commitment.

But the deeper problem is often simpler and more unsettling: we may have acted without a believable account of how change was supposed to happen in the first place.

That is the hidden value of theory based evaluation and theory of change thinking. Together, they challenge a familiar habit of modern institutions: confusing activity with causality. They force a harder question, not just “Did we do something?” but “What chain of events had to occur for this effort to work, and did those events actually occur?”

This is not a technical distinction. It is a philosophical one. It asks whether organizations are merely busy, or whether they are intelligent about change.

---

The real object being evaluated is not the program, but the logic behind the program

Most initiatives begin with a hopeful leap. A group identifies a problem, invents a solution, and turns that solution into an intervention. The logic is often implicit: if we provide training, people will gain skills; if people gain skills, they will behave differently; if behavior changes, outcomes will improve.

The problem is that these chains are rarely as linear as they look on paper. Real life is full of mediators, bottlenecks, timing effects, political constraints, cultural resistance, and unintended consequences. A program can be well delivered and still fail because one assumption in the causal chain was wrong. It can also appear to fail while quietly producing useful intermediate changes that were never measured.

This is why theory based evaluation matters. It shifts evaluation from a verdict on outputs to an inquiry into mechanisms. Instead of treating the program as a black box, it opens the box and asks what is happening inside. The central object of study becomes the logic of change, not merely the existence of an intervention.

A useful analogy is medicine. A doctor would not evaluate a treatment only by asking whether a pill was swallowed. The key question is whether the medication reached the right target, interacted with the body as expected, and produced the intended physiological response. If the symptom did not improve, the doctor does not automatically conclude the treatment was useless. The question becomes where the chain broke.

Development, public policy, philanthropy, and organizational change need that same discipline.

A program is not a theory. It is a bet on a theory.

That sentence changes everything. It means every initiative already contains an explanation of the world, whether written down or not. Evaluation should test that explanation, not merely record whether the initiative was implemented.

---

Theory of change is not a diagram. It is a disciplined argument about reality

Too many organizations reduce a theory of change to a slide with arrows. Boxes appear, arrows connect them, and the document is filed away as if the mere presence of a causal map guarantees clarity. But a genuine theory of change is not decoration. It is an argument.

That argument has at least four parts:

1. The problem definition: What is the exact condition we are trying to alter?
2. The causal pathway: By what sequence of changes should the intervention produce the outcome?
3. The assumptions: What must be true for each step in the pathway to hold?
4. The evidence plan: How will we know whether the assumptions and steps are actually occurring?

This structure matters because it makes disagreement productive. Instead of arguing vaguely about whether a program is “good” or “bad,” stakeholders can inspect the linkages. Maybe the intervention is sound, but the assumption about participant motivation is weak. Maybe the first mile of the pathway is strong, but the final mile depends on institutions that were never engaged. Maybe the intervention is excellent for a subset of people, but the theory falsely assumes a universal response.

A strong theory of change is therefore not a promise of success. It is a map of where success would have to come from. That makes it valuable even when it proves wrong, because a wrong theory, once exposed, is cheaper than a failed program that never learned why it failed.

Think about a job training initiative. The easy version says: train people, then jobs will follow. The better version asks whether employers actually trust the credential, whether transportation is a barrier, whether childcare limits attendance, whether the local labor market has openings, and whether the training aligns with hiring needs. Each of those is a causal link or assumption. If you do not make them explicit, you are not really designing change. You are hoping for it.

---

The hidden tension: evaluation wants proof, change is always probabilistic

Here is the deepest tension connecting these ideas. Institutions want certainty, but social change is rarely certain. Evaluation asks for evidence, but change emerges through complex systems in which no single factor controls the result.

This creates a trap. If we demand simplistic proof, we end up measuring only what is easy. Attendance, number of workshops, reports produced, dollars spent. These are visible and countable, but they are not the same as causal effectiveness. On the other hand, if we surrender to complexity, we risk becoming vague, treating any outcome as too messy to understand.

The solution is not to choose between rigor and realism. It is to use a causal mindset that is rigorous about uncertainty.

That means evaluation should not ask, “Did the intervention cause the outcome, yes or no?” in a naïve binary way. It should ask:

• What effects were expected at each stage?
• Which effects appeared, which did not, and in what sequence?
• What conditions strengthened or weakened the pathway?
• For whom did the theory work, and for whom did it break down?
• What alternative explanations remain plausible?

This is more than methodological finesse. It is humility with structure.

Imagine two school programs aimed at improving literacy. Program A gives every student one laptop. Program B changes reading instruction, engages parents, trains teachers, and adapts materials for struggling readers. If test scores do not rise immediately, Program A may still look impressive because it distributed visible equipment. Program B may look messy because it dealt with the less glamorous realities of teaching. A theory based evaluation asks which one changed the actual conditions for learning.

The difference is profound. One approach measures activity. The other measures whether the causal machinery was altered.

Complexity is not an excuse for confusion. It is a reason to become more explicit about assumptions.

---

A better mental model: from project thinking to ecosystem thinking

One way to deepen this conversation is to stop thinking of interventions as isolated projects and start thinking of them as disturbances in an ecosystem.

In an ecosystem, outcomes are not produced by a single heroic force. They emerge from interactions among many elements: actors, incentives, norms, institutions, feedback loops, and timing. Introduce one change, and the system adapts. Sometimes the response reinforces the intervention. Sometimes it neutralizes it. Sometimes it creates unintended side effects elsewhere.

This is why linear planning often disappoints. It assumes the world behaves like a machine with a few controllable levers. But many social systems behave more like weather patterns. You can influence them, but you do not command them.

A theory of change works best when it respects this reality. The goal is not to create a fake certainty. The goal is to identify the most fragile assumptions in the pathway and test them early. If a program depends on trust, build trust metrics. If it depends on institutional coordination, measure coordination. If it depends on sustained behavior change, look for evidence of habit formation, not just initial uptake.

This creates a practical discipline I call assumption sensitivity. Ask: if this intervention fails, which assumption is most likely to have been false? Which link in the chain is most vulnerable to context? Where would a skeptic attack the logic first? Those questions often reveal more than post hoc reporting ever can.

For example, a nutrition program may assume that families will use vouchers to buy healthier food. But if markets are distant, if fresh food spoils quickly, or if social norms discourage changing diets, then the theory is flawed, no matter how beautifully the program is delivered. The lesson is not that the program team was careless. It is that reality always has veto power over design.

The best theories of change therefore do not merely describe a route to success. They identify the conditions under which the route exists at all.

---

Evaluation becomes more useful when it behaves like inquiry, not judgment

Many people fear evaluation because they associate it with punishment. In practice, this fear produces defensive behavior. Teams choose safe metrics, overstate certainty, and avoid learning questions that might expose weakness. But theory based evaluation is at its best when it functions less like a courtroom and more like a laboratory.

A laboratory does not ask whether a hypothesis deserves moral approval. It asks whether the hypothesis survives contact with evidence. That shift changes organizational culture. People become more willing to surface weak links because weakness is no longer treated as failure of character. It becomes information.

This is especially important in complex, multi actor efforts such as community safety, climate adaptation, public health, or international development. In these settings, no one actor controls the outcome. Success depends on alignment across institutions, incentives, and public behavior. A useful evaluation therefore asks not only whether results improved, but where the system became more or less capable of producing those results in the future.

This is one reason theories of change are so valuable in collaborative work. They give different stakeholders a common language for disagreement. One group may care about service delivery, another about policy reform, another about community trust. The theory of change can hold these together by making visible how each piece contributes to a larger causal chain.

Without that shared logic, collaboration often becomes a coalition of preferences. With it, collaboration becomes a testable account of interdependence.

---

The practical payoff: better decisions before, during, and after action

The true power of theory based evaluation is that it improves decision making at every stage.

Before action, it forces clarity. If you cannot explain how the intervention should work, you probably should not scale it.

During action, it creates feedback. If early assumptions fail, you can adapt rather than continue blindly.

After action, it enables learning. If outcomes were weak, you can identify where the pathway broke instead of declaring the whole effort a waste.

This is much more than a management tool. It is a way of respecting reality. It prevents the common error of confusing a plausible story with a verified one. It also helps organizations resist the temptation to overclaim. Many programs are not fully successful or fully unsuccessful. They are partially right about some assumptions, wrong about others, and dependent on context in ways they did not anticipate. A theory based lens can capture that nuance.

Consider a youth mentorship program. The simple metric might be graduation rates after three years. But a stronger theory of change might track whether participants build adult trust, increase attendance, improve self regulation, and expand postsecondary aspirations. If graduation rates do not move immediately, those intermediate effects may still matter because they tell you whether the pathway is strengthening. If none of the intermediate effects appear, the problem is not timing. It is the logic itself.

That distinction saves time, money, and morale.

---

Key Takeaways

1. Treat every program as a bet on a causal theory. If you cannot name the theory, you cannot meaningfully evaluate the result.
2. Map the full pathway, not just the outcome. Identify inputs, intermediate changes, assumptions, and context, not only the final target.
3. Test the fragile links first. Focus early attention on the assumptions most likely to fail in the real world.
4. Look for partial effects and intermediate signals. A program may be working in ways that final outcome metrics do not yet reveal.
5. Use evaluation to learn, not just to judge. The best evaluation improves the next decision, not merely the last report.

---

The deeper lesson: change is earned through logic before it is visible in results

The biggest mistake in improvement work is to believe that outcomes are the primary proof of intelligence. They are not. Outcomes are the late stage of a process. Long before a result appears, there has to be a credible explanation for how action becomes transformation.

That is why theory of change is more than a planning exercise and theory based evaluation is more than an auditing tool. Together they represent a mature stance toward reality. They say: if we want to change the world, we must first understand the world’s causal grammar.

And once you see that, the meaning of success changes. Success is no longer just getting a good number at the end. Success is building an intervention that can explain itself, adapt to evidence, and survive contact with actual human systems.

In that sense, the best programs are not the ones with the loudest claims. They are the ones with the clearest theories, the sharpest assumptions, and the courage to let reality revise the story.

That is the real test. Not whether we acted. Not even whether we hoped well. But whether we knew, with enough precision to be tested, how change was supposed to happen at all.