The Discipline of Noticing What You Did Not Plan

What if the real problem is not bad plans, but blind plans?

Most organizations are excellent at drawing maps of the future and surprisingly poor at reading the terrain once they begin walking. They define goals, theories, milestones, indicators, and assumptions, then treat the resulting plan as if it were a reliable photograph of reality. But reality rarely behaves like a photograph. It behaves more like weather: shifting, localized, and often indifferent to our carefully prepared forecasts.

That creates a deeper question than evaluation or planning alone can answer: how do you know what is actually happening when the world does not follow your model? The usual answer is to check whether the plan worked. Yet that answer is often too narrow. It assumes the most important thing to observe is whether predefined outcomes appeared on schedule. In practice, the most valuable learning often comes from what no one expected, what no one measured, and what no one initially believed mattered.

The most important evidence is often not the evidence you designed your system to see.

This is the hidden connection between outcome oriented evaluation and theory of change thinking. One discipline asks us to look for meaningful change wherever it appears. The other asks us to make the assumptions behind our expected change explicit. Together, they reveal a better way to think about strategy: not as a blueprint to execute, but as a living hypothesis to test, revise, and occasionally abandon.

---

The trap of elegant certainty

A theory of change is powerful because it makes causal thinking visible. It says, in effect, “If we do X, then Y should happen because A, B, and C are true.” That structure is immensely useful. It forces teams to surface hidden logic, name dependencies, and articulate the conditions under which an intervention might succeed. In a world of vague aspirations, that clarity is a gift.

But clarity can quietly become a trap. Once the theory is written down, teams often begin to protect it. They gather evidence that confirms it, design indicators that match it, and celebrate when the numbers move in the expected direction. Over time, the theory of change can turn from a learning tool into an identity statement: this is what we believe, therefore this is what we must prove.

That is where assumptions matter. Assumptions are not footnotes. They are the load bearing beams of the entire structure. If they are false, incomplete, or outdated, the whole logic of the plan can remain impressive on paper while failing in practice. A program aimed at increasing school attendance, for example, may assume that parents keep children home mainly because of low motivation, when the real constraint is unsafe transport or seasonal labor demands. The intervention may still be beautifully implemented, but it is aimed at the wrong causal story.

The key insight is that assumptions are not merely things to document. They are things to interrogate continuously. A serious strategy does not ask, “Did we follow the plan?” first. It asks, “Which assumptions held, which did not, and what did the world do that our model did not anticipate?”

---

Why outcomes are more interesting than indicators

Traditional measurement tends to privilege predefined indicators. Did enrollment rise? Did revenue increase? Did health outcomes improve? Such questions are useful, but they are also limiting because they only detect changes we already knew how to name. They are like fishing with a net designed for one species and concluding the lake is empty if that species does not appear.

A more generative approach begins with outcomes in the broad sense: observable changes in behavior, relationships, practices, policies, capacity, or condition. The point is not to worship novelty. The point is to stay open to the possibility that meaningful change may take an unexpected form. A new collaboration between agencies may matter as much as a new regulation. A shift in language among community leaders may matter more than a formal report. A small change in who speaks in meetings may signal a deeper redistribution of power.

This is where the tension becomes productive. If assumptions tell us what we expected to happen, outcome oriented inquiry tells us what actually happened, even when it does not fit the expected category. In other words, the first helps us avoid self deception; the second helps us avoid measurement blindness.

Think of it this way: a theory of change is a hypothesis about causality, while outcome harvesting is a search for evidence of change in the wild. Hypotheses need structure. Searches need openness. When combined, they create a more intelligent learning loop than either one can produce alone.

Imagine a foundation funding youth employment initiatives. Its theory of change assumes that training leads to better jobs. But outcome oriented observation might reveal something else: graduates form informal peer networks that help them access contracts, not salaried employment. If the team is only looking for the original outcome, it may judge the program as mediocre. If it is looking for real world change, it may discover a different and potentially more effective pathway.

That is the deeper lesson: the world often answers the question you should have asked, not the one you wrote down.

---

A better model: plans as hypotheses, outcomes as signals

The most useful way to combine explicit assumptions with outcome oriented learning is to stop treating plans as commitments to specific results and start treating them as provisional hypotheses about how change might unfold.

Here is a simple mental model:

1. Assumption layer: What must be true for the intervention to work?
2. Mechanism layer: Through what pathway do we expect change to happen?
3. Signal layer: What kinds of real world changes would suggest the mechanism is operating, even if they are not our original targets?
4. Revision layer: What do we update when the signals contradict the assumptions?

This framing is powerful because it preserves discipline without sacrificing adaptability. It says we should not abandon planning, but we should plan in a way that expects surprise. Instead of asking only whether the intervention hit the target, ask whether the underlying causal story still makes sense.

Consider a nonprofit supporting local entrepreneurship. Its initial theory might assume that access to microcredit is the bottleneck. But outcome signals may reveal that the real barrier is unstable household income, which makes repayment psychologically and financially risky. Another signal might show that women entrepreneurs gain more from mentorship and trusted market connections than from loans. In that case, the original assumption was not useless. It simply pointed toward a causal relationship that turned out to be incomplete.

This is where many organizations get stuck. They either cling to the original model or throw it away too quickly. The better move is to distinguish between three things:

• Broken assumption: a belief that the world has disproved
• Incomplete assumption: a belief that is partly true but missing key conditions
• Emergent assumption: a new pattern that becomes visible only after action begins

Once you see those distinctions, evaluation stops being a verdict and becomes a conversation.

---

The real work is epistemic humility

Underneath both approaches is a deeper virtue: epistemic humility, the willingness to admit that our understanding of change is always partial.

That does not mean lowering standards. It means raising them. A humble organization does not settle for vague intuition. It takes assumptions seriously enough to name them, test them, and discard them when needed. It also takes outcomes seriously enough to notice evidence that falls outside the plan. The combination is rare because it requires both structure and openness, two qualities that organizations often mistakenly treat as opposites.

In practice, humility looks like this:

• Asking frontline staff what is changing before asking whether the dashboard moved
• Looking for unintended outcomes, not just intended ones
• Treating stakeholder stories as data, not decoration
• Rewriting the theory of change when reality supplies better causal clues
• Distinguishing between implementation failure and theory failure

A useful analogy is gardening. A gardener can prepare soil, choose seeds, set irrigation, and schedule care. But no amount of planning makes the plant grow in a straight line according to the gardener’s PowerPoint. Sunlight, pests, weather, soil chemistry, and neighboring plants all matter. Good gardening is not passivity. It is attentive intervention, guided by both intention and observation.

Many organizations need to become better gardeners of change.

A strong theory of change does not predict the future perfectly. It teaches you how to notice when the future is becoming different from what you imagined.

That shift in mindset changes everything. It moves teams away from defensive evaluation, where evidence is collected to defend a program, toward generative evaluation, where evidence is collected to improve the program and the model behind it.

---

What this means for strategy, evaluation, and learning

If you combine these ideas seriously, several practical implications follow.

First, strategy should be written in a way that invites disconfirmation. The more explicit the assumptions, the easier it is to learn. Vague promises are hard to challenge, and therefore hard to improve. A well formed theory of change should specify not just what is expected to happen, but what would make that expectation wrong.

Second, evaluation should search beyond predefined outputs. Outputs tell you whether activities happened. Outcomes tell you whether the world changed. But the most valuable learning often sits in the gap between the two. That gap is where people improvise, adapt, resist, repurpose, or create side effects that eventually become the main effect.

Third, anomalies are not noise by default. Sometimes they are measurement error, but sometimes they are the first visible sign of a more interesting truth. If one region unexpectedly outperforms, or one beneficiary group changes in a surprising way, the task is not to dismiss the anomaly too quickly. The task is to ask what assumption it is testing.

Fourth, the goal is not to eliminate uncertainty, but to make uncertainty usable. Organizations often spend enormous energy trying to reduce ambiguity before acting. Yet change work is inherently uncertain. Better to build systems that can sense, adapt, and revise in motion than to wait for perfect certainty that never arrives.

Finally, the most valuable learning often comes from outcomes nobody originally intended. A youth program may not produce jobs immediately, but it may create confidence, networks, or civic participation. A health initiative may not shift one metric, but it may change trust in institutions. If those effects matter, they belong in the story of value.

---

Key Takeaways

1. Treat your theory of change as a hypothesis, not a creed. Write down what must be true for it to work, then revisit those beliefs regularly.

2. Look for outcomes, not just indicators. Indicators confirm what you already measured. Outcomes reveal what changed, including unexpected changes.

3. Separate implementation failure from assumption failure. A weak result does not always mean poor execution. Sometimes the causal logic was incomplete or wrong.

4. Make anomalies discussable. Create a habit of asking, “What might this surprising result be telling us?” instead of dismissing it.

5. Use learning loops, not one time evaluations. Build review points into the work so that each round of action sharpens the next theory.

---

Conclusion: from proving change to perceiving change

The deepest mistake in strategy is not overconfidence. It is premature certainty. We often assume that because we can articulate a pathway to change, we can control it. But change is not a line drawn on paper. It is an evolving system of relationships, incentives, behaviors, and constraints that only partly reveals itself in advance.

That is why the most powerful organizations are not the ones with the most elegant plans. They are the ones that can hold a plan lightly, observe reality carefully, and revise intelligently. They understand that assumptions are not embarrassing admissions, but necessary starting points. They understand that outcomes are not just targets to hit, but signals to interpret.

In the end, the point is not to choose between planning and discovery. The point is to build a practice that can do both. When you name your assumptions honestly and harvest outcomes attentively, you stop asking only whether the world obeyed your plan. You start asking the more useful question: what is the world trying to teach us about how change really happens?