Why Good Theory of Change Models Fail When They Stop Treating Assumptions as Hypotheses

The seductive lie of the tidy plan

A theory of change can look beautifully persuasive on paper. Boxes line up. Arrows point in the right direction. Outcomes appear to follow interventions with almost moral certainty. And yet the most dangerous thing about a well drawn map is that it can make us forget the terrain is alive.

That is the core tension at the heart of every serious theory of change: we want causality to feel stable, but reality behaves conditionally. Programs do not unfold in a vacuum. People adapt, institutions resist, incentives shift, and context quietly rewrites the script. A model can look rigorous while hiding its most important weakness, which is not the absence of logic, but the presence of unexamined assumptions.

This is why the real value of a theory of change is not that it predicts the future. Its value is that it exposes what must be true for a desired future to happen at all.

A theory of change is not a promise. It is a bet, with conditions attached.

That distinction matters. The best models do not pretend to eliminate uncertainty. They make uncertainty visible enough to test, update, and learn from. The moment a theory of change is treated as a finished explanation instead of a living hypothesis, it stops being useful and starts becoming decorative.

---

The hidden architecture of action

Most planning documents confuse sequence with causation. They assume that if step A happens, then step B will follow, and finally outcome C will arrive. But in practice, a theory of change is less like a row of dominoes and more like a suspension bridge: the visible structure matters, but its strength depends on hidden load bearing cables.

Those cables are assumptions. They are the unstated beliefs about behavior, resources, timing, legitimacy, trust, political will, and social norms that make the pathway plausible. If a literacy program assumes teachers will adopt a new method, the theory of change depends not only on training, but on whether teachers have time, support, incentives, and confidence to use it. If a public health intervention assumes families will respond to information, it also depends on whether the message is trusted, culturally legible, and practically actionable.

This is where many theories of change become fragile. They describe the what and sometimes the how, but not the under what conditions. That omission is not minor. It changes the nature of the model from a tool for learning into a tool for reassurance.

A strong theory of change does three things at once:

1. It clarifies the intended pathway from action to outcome.
2. It identifies the assumptions that could break that pathway.
3. It tells you what evidence would reveal whether those assumptions are holding.

When these three elements are present, the theory of change becomes an instrument of inquiry rather than a mural of intentions.

Consider a job training initiative. A shallow model says: train people, then they get jobs. A deeper model asks: do local employers recognize the credential, do participants have transportation, are the jobs compatible with caregiving responsibilities, and does the labor market actually need the skills being taught? The last questions are not side notes. They are the mechanism by which the intervention succeeds or fails.

In other words, assumptions are not decorative caveats. They are the gears inside the machine.

---

Why assumptions are the real theory

There is a subtle but profound inversion here: the formal diagram is often the least interesting part of a theory of change. What matters most is the logic of the assumptions, because assumptions reveal how the designer imagines reality works.

Every assumption is really a miniature theory about human behavior or system dynamics. It says, implicitly, that people will do X if Y happens, that institutions will respond to incentive Z, or that context will remain stable enough for the pathway to hold. This is why two programs with identical activities can produce opposite results. They may be operating under different, unspoken theories of human and organizational change.

Think of assumptions as the risk surface of a model. The more ambitious the intervention, the more assumptions it must cross before reaching the outcome. A small program may need only one or two conditions to hold. A systems change initiative may depend on dozens. The problem is not that ambitious theories of change are bad. The problem is that ambition often disguises fragility.

A useful mental model is to distinguish between three kinds of assumptions:

1. Behavioral assumptions

These concern whether people will notice, trust, understand, and act. Will participants show up? Will staff adopt the method? Will decision makers care enough to change policy?

2. Structural assumptions

These concern the broader environment. Are the necessary resources available? Are legal, political, and institutional conditions supportive? Is the timeline realistic relative to bureaucracy or market cycles?

3. Interpretive assumptions

These are often overlooked. They concern whether stakeholders interpret the intervention the way the designer expects. A community may see a program as supportive, paternalistic, irrelevant, or threatening, all of which radically alter its effects.

Most failures happen not because an intervention is entirely wrong, but because one of these assumption layers was never made explicit. The model looked coherent, but it was only coherent inside the designer’s head.

The most important question in any theory of change is not “What do we want to happen?” It is “What must people and systems believe, value, and permit for this to happen?”

That question shifts planning from wishful sequencing to disciplined realism.

---

The best theories of change are falsifiable, not flattering

A theory of change should be judged by how well it helps us learn, not by how inspiring it sounds. That means the strongest models are those that can be wrong in specific ways. This is uncomfortable, because many organizations prefer theories that affirm their mission while politely avoiding sharp edges.

But a theory of change that cannot be challenged is not really a theory. It is branding.

To make a theory of change genuinely useful, convert assumptions into testable hypotheses. For example:

• Instead of saying, “Communities will engage,” specify, “At least 60 percent of invited participants will attend two or more sessions if meetings are held at accessible times and childcare is provided.”
• Instead of saying, “Teachers will adopt the method,” specify, “Teachers will use the new practice weekly when the training includes classroom coaching and principal support.”
• Instead of saying, “Policy change is possible,” specify, “The coalition will secure at least one champion inside the agency if evidence is packaged in a way that aligns with current political priorities.”

These are not just better metrics. They are better theories.

A falsifiable theory of change also changes how you respond to failure. If outcomes do not materialize, the question is not immediately, “Did the intervention fail?” The better question is, “Which assumption did reality reject?” That distinction prevents the all too common habit of blaming implementation for what may actually be a design problem, or blaming the design for what may actually be a context problem.

This is where theories of change become strategically powerful. They help distinguish among three possibilities:

1. The mechanism is wrong.
2. The mechanism is right, but the assumptions were false.
3. The mechanism is right, but the timing or dosage was insufficient.

Without that distinction, organizations often make the worst possible move: they intensify a broken model.

Imagine pouring more fuel into a car that has the wrong kind of engine. More effort does not fix the mismatch. It only makes the failure more expensive.

---

A practical framework: from blueprint to learning system

The future of theory of change work is not more complexity for its own sake. It is more epistemic humility built into the design process. In practice, that means treating the model as a learning system with feedback loops, not as a one time planning artifact.

Here is a simple framework that can make a theory of change more useful immediately.

Step 1: Separate the pathway from the assumptions

Write the causal sequence first, then list every assumption needed at each step. Do not bury assumptions in footnotes or narrative prose. Make them visible.

For example:

• Activity: provide mentoring.
• Expected effect: participants gain confidence and skills.
• Assumptions: mentors are consistent, participants trust mentors, meetings are frequent enough, and the skills are relevant to the participants’ actual barriers.

This separation prevents the diagram from doing emotional work that should be done by evidence.

Step 2: Rank assumptions by fragility

Not all assumptions deserve equal attention. Some are routine. Others are high leverage and high risk. Ask two questions:

• If this assumption is false, does the whole pathway collapse?
• How uncertain are we that it is true?

The intersection of high impact and high uncertainty is where validation should start.

Step 3: Design early tests, not just endline measures

Waiting until the end to discover whether an assumption was false is too late. Build in early signals.

For instance, if a civic engagement program assumes people will speak honestly in public meetings, measure whether attendance is diverse, whether quieter participants are heard, and whether the format is psychologically safe before measuring downstream policy influence.

Step 4: Use negative evidence as intelligence

If a theory of change was wrong, that is not embarrassment. It is information about the system. The goal is not to prove the model right. The goal is to find the model that best fits reality.

Step 5: Revisit the model when context changes

A theory of change ages. Political transitions, economic shocks, staffing shifts, and cultural changes can all invalidate assumptions that once held. The model should be reviewed as a living document, especially when the environment moves.

This framework turns planning into a cycle: hypothesize, test, learn, revise.

That cycle is more honest than pretending the future can be charted once and for all.

---

What becomes possible when we stop worshipping the diagram

The deepest benefit of this approach is not better paperwork. It is better judgment.

When organizations stop treating the theory of change as a static explanation, several things improve at once. Teams become more precise about why they believe an intervention should work. Funders become better able to compare programs based on the quality of their causal reasoning, not just the polish of their presentation. Practitioners become more alert to contextual drift. And learning becomes faster because failures are interpreted as information rather than noise.

There is also a cultural shift. A mature organization becomes less attached to appearing right and more committed to becoming less wrong. That sounds modest, but it is a profound advantage in complex environments.

The same logic applies beyond nonprofits and policy. In business, strategy often fails when leaders mistake assumptions for facts. In education, curricular reforms fail when they ignore the conditions teachers need in order to teach differently. In technology, products fail when designers confuse user intent with user behavior. In each case, the issue is not the absence of a plan. It is the presence of an untested story about how change happens.

A theory of change, at its best, is a discipline of attention. It asks you to notice the hidden predicates of success, the conditions under which your ambitions become plausible. That attention is rare. It is also indispensable.

The real power of a theory of change is not in telling you where the world should go. It is in revealing what the world must already be ready to do for change to stick.

---

Key Takeaways

1. Treat assumptions as the core of the model. If you cannot name the conditions that make an intervention work, you do not yet have a usable theory of change.

2. Convert hopes into hypotheses. Rewrite vague beliefs such as “people will engage” into specific, testable claims with observable indicators.

3. Rank assumptions by fragility and importance. Focus first on the assumptions that are both uncertain and essential to the pathway.

4. Test early, not just at the end. Build in small experiments or leading indicators that reveal whether the causal pathway is holding before full scale rollout.

5. Revise the model when reality changes. A theory of change should evolve as context changes, not remain frozen as a symbolic artifact.

---

The question worth asking next

The biggest mistake we make with theories of change is believing they are about control. They are not. They are about contact with reality.

A good theory of change does not reassure us that success is guaranteed. It makes us more honest about what success depends on. And that honesty is not a limitation. It is the beginning of serious work.

So the next time a model looks elegant, do not ask first whether it is inspiring. Ask whether its assumptions are alive, visible, and testable. Because in the end, the future does not reward the prettiest diagram. It rewards the clearest understanding of what must be true for change to happen.