Why Good Change Fails Without a Theory of What to Notice

The Hidden Problem in Most Change Efforts

Most change efforts fail for a strange reason: they are designed to prove an idea before the world has had a chance to answer it. We set targets, define milestones, and assign ownership as if reality were a straight corridor. But social systems, organizations, communities, and ecosystems do not behave like corridors. They behave like weather: dynamic, nonlinear, and full of feedback loops that make yesterday’s success a poor guide to tomorrow’s result.

That creates a deep tension. We want change to be intentional, but the more complex the setting, the less useful it is to pretend we can fully predict outcomes in advance. The real question is not, “Can we control the future?” It is, “How do we learn our way into the future without fooling ourselves?”

This is where many strategies break down. They confuse a theory of change with a prediction, and they confuse evidence with indicators that were preselected before anything meaningful happened. In complex settings, that is often the wrong sequence. If the world changes in response to our interventions, then the most important evidence may be the evidence we did not know to look for at the beginning.

In complex change, the central task is not to execute a plan perfectly. It is to notice what the plan could not have anticipated.

---

Complexity Changes the Meaning of “Success”

In a simple system, cause and effect behave politely. If you put a key in a lock, the door opens. In a complex system, the same action can produce different effects depending on timing, relationships, history, and context. A training program may improve performance in one department and quietly fail in another. A policy may trigger compliance on paper while generating resistance in practice. A community intervention may shift norms in one neighborhood and leave another untouched.

This means success cannot be reduced to whether a prewritten target was hit. That kind of thinking assumes the future is already known, and that the only question is whether implementation was faithful. But in complex environments, the intervention itself changes the conditions of observation. People adapt, institutions respond, and new problems emerge. The system is not a machine being repaired. It is a living field of interaction.

A useful mental model is to distinguish between complicated and complex problems. Complicated problems, like building a bridge, require expertise and coordination. Complex problems, like reducing violence or improving governance, require learning, adaptation, and interpretation. The bridge has a blueprint. The social field has a conversation.

Once that distinction is clear, the logic of evaluation must change. Instead of asking only, “Did we get the intended result?” we also need to ask, “What happened that mattered, and how do we know?” This shift may sound subtle, but it is decisive. It moves change work from mechanical verification toward disciplined attention.

---

Outcome Is Not the Same as Evidence

One of the most important mistakes in change work is treating outcomes as if they are self-explanatory. They are not. An outcome is a change in the world. Evidence is the trail that lets us understand how that change relates to action, context, and influence.

This matters because complex change rarely announces itself in a neat line from input to output to outcome. The things that matter often appear as fragments: a policy draft cited in a new meeting, a farmer adapting a practice after a peer exchange, a health worker informally changing how she speaks to patients, a local group adopting a language that did not exist in the original proposal. These are not always the headline indicators, but they may be the first signs that a larger shift is taking place.

An outcome harvesting mindset begins with the outcome and works backward to identify what contributed to it. That reverses the usual logic of evaluation. Instead of asking, “Did we cause the predefined change we expected?” it asks, “What meaningful changes occurred, and what evidence connects them to the work?” This is especially powerful when the change process is emergent, distributed, or politically contested.

Consider a civil society coalition trying to influence a city’s housing policy. If the coalition only tracks whether a law passed, it may miss the more important story. Perhaps the coalition changed the language officials now use, altered who gets invited into planning conversations, and reshaped public expectations long before any law was enacted. Those shifts are not secondary. They may be the very mechanism through which policy eventually changes.

In complex systems, the visible outcome is often only the last scene in a much longer chain of invisible adjustments.

This is why evidence must be treated as a living inquiry, not a box to tick. The point is not merely to count what happened. The point is to discover what the system is becoming.

---

A Better Theory of Change: From Line to Loop

Traditional theories of change often look like arrows on a slide: activity leads to output, output leads to outcome, outcome leads to impact. The problem is not that these diagrams are useless. It is that they are incomplete in the places where reality is most interesting.

A better theory of change for complex settings should look less like a line and more like a loop. It should include at least four elements:

1. Intent: what we are trying to influence.
2. Signals: what changes we expect to notice if influence is occurring.
3. Feedback: how the system responds to those signals.
4. Revision: how we adjust our assumptions based on what we learn.

This framework changes the role of planning. Planning is no longer a one time declaration of certainty. It becomes a hypothesis about where attention should be directed. That is a much humbler, and much more useful, function.

Imagine a public health team working to increase vaccination uptake. A linear model might focus on vaccine doses delivered and appointment numbers completed. A loop based model would also watch for community narratives, trust shifts, misinformation patterns, and changes in the behavior of local messengers. If uptake rises, the team still has to ask: Did trust improve, did access become easier, did a respected leader endorse the program, or did people simply comply because of a requirement? Different mechanisms imply different next steps.

This is where complexity and outcome harvesting meet in a powerful way. Complexity tells us not to overclaim causality in advance. Outcome harvesting gives us a disciplined way to search for meaningful change after the fact, even when that change was not predicted.

Together they suggest a new principle: a theory of change should not only describe what we hope will happen. It should describe what we will be alert to if reality begins to move in an unexpected direction.

That is a profound shift. It means the quality of a theory of change is measured not just by how elegant it looks, but by how well it trains attention.

---

The Real Skill Is Not Prediction, but Sensible Noticing

If this all sounds abstract, think about how experienced doctors work. The best clinicians do not rely solely on checklists or initial diagnoses. They observe patterns, watch for deviations, compare signals over time, and revise their understanding as more information appears. They do not treat uncertainty as failure. They treat it as the normal condition of responsible judgment.

Change work in complex systems should be approached the same way. The task is not to eliminate uncertainty. The task is to build a practice of sensible noticing.

Sensible noticing has three features.

First, it is open. It can register unexpected changes rather than only confirming predefined expectations. If a program designed to improve job placement instead reveals stronger peer networks among participants, that should count as a meaningful signal, not a distraction.

Second, it is disciplined. Open does not mean vague. A good harvesting process still asks for evidence, attribution logic, and plausible contribution. It does not accept anecdotes at face value. It tries to understand why a change matters and what else might explain it.

Third, it is adaptive. The evidence gathered should feed back into strategy. If the system is resisting one pathway, perhaps another pathway is emerging. If an intervention is producing a side effect, perhaps that effect reveals a deeper leverage point.

A concrete example helps. Suppose an education initiative introduces a new reading method in several schools. A linear evaluator might ask whether test scores went up. A sensible noticing approach would also look for changes in teacher collaboration, parental engagement, student confidence, and local adaptation of materials. Maybe the biggest shift is not immediate scores, but the fact that teachers begin sharing lesson plans across schools. That network effect may be the real engine of future improvement.

This way of working does not lower standards. It raises them. It demands that we pay attention to the right level of reality. In complex systems, the right question is often not, “Did the intended number move?” but, “What changed in the system’s behavior, relationships, and language?”

---

A Practical Framework for Complex Change Work

If the world is more like weather than a machine, how should we act? Here is a simple framework that combines strategic clarity with adaptive learning.

1. Name the change you care about, but do not over specify the route

Be clear about the direction of travel. But resist the temptation to map every step in advance. In complex contexts, specificity about destination is more valuable than false certainty about process.

2. Define the signals that would tell you the system is shifting

Do not rely only on final indicators. Identify intermediate signs: new language, new relationships, new decision habits, new alliances, new forms of resistance, new norms.

3. Harvest outcomes continuously

Do not wait until the end to ask what changed. Build regular practices for collecting meaningful change stories, corroborating them, and tracing contribution. This helps you spot patterns before they harden into missed opportunities.

4. Treat surprises as data

Unexpected outcomes are not noise to be filtered out. They may be the system telling you something important. A surprising alliance, a failed pilot that spreads in a modified form, or an unintended behavior shift may reveal where leverage really lies.

5. Revise the theory, not just the tactics

If evidence consistently points in a different direction, do not only tweak implementation. Reexamine the underlying assumptions. Perhaps the issue was never access, but trust. Perhaps the barrier was never information, but status. Perhaps the mechanism of change is not persuasion, but imitation.

This framework is useful because it acknowledges a basic truth: in complex systems, the most valuable intelligence often arrives after action begins. That does not make planning pointless. It makes planning provisional.

---

Key Takeaways

• Stop treating theories of change as predictions. In complex settings, they are better understood as hypotheses about what to watch for and how to learn.
• Look for meaningful change, not only planned change. Some of the most important outcomes are emergent, indirect, or unexpected.
• Use evidence to trace contribution, not to oversimplify causality. The goal is to understand how an intervention interacted with context, not to claim total credit.
• Build feedback loops into strategy. Let what you discover change what you do next, including your assumptions.
• Train teams to notice signals early. New language, relationships, norms, and behaviors often reveal more than endline metrics.

---

The Reframe: Change Is a Conversation With Reality

The deepest insight here is not about evaluation technique. It is about humility. In complex systems, change is not a matter of imposing a plan on passive material. It is a conversation with reality, one in which reality speaks back through outcomes, surprises, side effects, and partial shifts.

That means the most intelligent change agents are not the ones who predict best. They are the ones who notice best, learn fastest, and revise honestly. They understand that a theory of change is only as good as its capacity to be tested by the world, and improved by what the world reveals.

So perhaps the right question is not, “Did we follow the plan?” The better question is, “What did the system tell us, and what are we now able to see that we could not see before?”

That reframe changes everything. It turns evaluation from a courtroom into a laboratory, and strategy from a script into a search.

In the end, good change work is less about proving that we were right. It is about becoming more precise in how we pay attention to what becomes possible.