A Leader’s Framework for Decision Making

A decision fails in three distinct places, and only one of them gets discussed. The first is the situation itself — whether cause and effect are tightly linked, loosely linked, or not linked at all. The second is the leader's reading of that situation. The third is the method the leader picks in response. Fit is the alignment of all three. When they line up, the response works because it was built for the terrain it landed on. When they don't, the response can be flawless in execution and still produce the wrong result.

The common assumption is that a good decision is a matter of skill or resolve. It is closer to a matter of matching. A tightly ordered problem rewards analysis and best practice; a genuinely complex one punishes them, because the pattern only becomes visible after you act. The same decisive move that restores order in a crisis will suffocate a situation that needed room to reveal itself. There is no universally strong response, only a response that fits or does not.

The diagnosis is the hinge, and it is the step most often skipped. Leaders tend to reach for the method they trust rather than the one the situation calls for, which is how a correct read gets paired with a wrong tool, or a wrong read gets paired with a tool that would have been perfect for the situation the leader imagined instead of the one in front of them.

Fit is also fragile in a specific way: it degrades under the pull of habit and expertise. The domains where a leader is most fluent are the domains where misreading is most likely, because fluency substitutes for looking. High fit is less an achievement than a discipline of checking that the terrain is what you assumed before you commit the response you already prefer.

How to

1. After acting, check all three links: was the context what you thought, did your diagnosis match it, and did your method match your diagnosis?
2. Watch for domain traps that degrade fit even after a correct diagnosis, and adjust the response before results deteriorate.
3. Treat persistent poor outcomes as a signal to re-examine which of the three links is broken, rather than pushing harder on the same method.

Watch out for

• Assuming that because your diagnosis was right, any downstream failure must be an execution problem.
• Locking in a method after achieving fit, when the underlying context has quietly shifted beneath it.

Effectiveness is not one quality but several, and they do not travel together. A leader can avoid obvious mistakes and still miss the opening for something new. Another can restore order in a crisis and then apply the same forceful grip to a settled situation that needed patience instead. The measure of a decision is whether it produced the result the situation actually permitted — order where order was possible, discovery where discovery was the point, damage avoided where the downside was real.

This is why effectiveness cannot be judged from the outcome alone. A decision that turned out well may have been a poor match for its context that got rescued by luck. The more durable signal is whether the response fit the situation: when the reading of the terrain, and the method chosen for it, align with how cause and effect really behave, good results follow with far more regularity than chance would give.

The steady drag on effectiveness comes from the familiar. A leader's strongest domains are where confident, practiced responses feel most obviously right — and where the wrong response is easiest to deploy without noticing, because it arrives dressed as experience. Effectiveness, then, is less about producing brilliance and more about not letting a well-worn instinct answer a question it was never suited to.

How to

1. Define what success looks like per context — error avoidance in simple, sound analysis in complicated, useful emergence in complex, restored stability in chaotic.
2. Evaluate decisions on process-context match, not just outcomes, especially where outcomes are partly luck.
3. Separate the review of a failed probe (did we learn?) from the review of a failed simple decision (did we ignore best practice?).

Watch out for

• Punishing safe-to-fail experiments that fail, which teaches your team to stop probing exactly where probing is needed.
• Crediting a good outcome that arose from a context-mismatched decision, hardwiring a habit that will eventually break.

New business models and genuinely creative solutions do not arrive on schedule, and they cannot be ordered up by analysis. They surface, which is a different verb. In situations where the connection between action and result only becomes clear in hindsight, the way to find an opportunity is to run small experiments that are safe to fail — probes that either die cheaply or reveal a pattern worth amplifying. The value is not in any single experiment but in acting your way into knowledge that could not have been reasoned out in advance.

When a situation has broken down into disorder, the same instinct takes a different shape. The pressing task is to reestablish footing, but the moment also opens room that stability never allows. Running innovation efforts in parallel — one line of work stabilizing, another probing for what the upheaval has made newly possible — lets a leader capture emergent opportunity without waiting for calm to return.

What governs all of this is whether the response fits the context. Safe-to-fail experiments are wasted on a problem with a known answer, and best-practice analysis is useless against a situation whose rules haven't formed yet. Discovery follows the match. A leader who reads the terrain correctly and reaches for the method that suits it will keep finding openings others walk past — not through vision, but through using the instrument the moment was built for.

How to

1. In complex contexts, run multiple safe-to-fail experiments in parallel, designed so that failure teaches you something and success can be scaled.
2. In chaotic contexts, stand up a separate team to hunt for opportunity while the main effort restores order.
3. Set probes that are diverse rather than convergent, so you explore a range of the possibility space rather than variations on one bet.

Watch out for

• Killing instability too fast and eliminating the very conditions from which new opportunities emerge.
• Designing 'experiments' that cannot actually fail, which yields confirmation rather than discovery.

The first decision a leader makes is not what to do but where they are standing. Before any response can be right, the situation has to be read correctly, and the reading turns on a single question: how does cause connect to effect in front of you right now. In some situations that link is clear and holds still. In others it can be worked out, but only by someone who knows the field. In still others it reveals itself only after the fact, and in a few it cannot be traced at all.

Those differences sort a situation into one of five contexts: simple, complicated, complex, chaotic, or the null state of disorder, in which no one yet knows which of the other four applies. Diagnosis is the act of naming which one you are in. It is not a personality assessment or a mood; it is a claim about the structure of the problem itself.

The cost of skipping this step is that leaders reach for the response they are most comfortable with rather than the one the situation requires. A method built for a stable, ordered problem will fail in an unstable one, and vice versa, and the failure will not announce itself as a diagnostic error. It will look like bad execution. Getting the context right is what makes everything downstream possible, because the correct response method follows directly from it.

Disorder deserves its own respect. Most of the danger sits there, because leaders in disorder do not know they are in it, and they fall back by default on whatever style they already prefer. Recognizing that you cannot yet read the situation is itself a diagnosis, and often the most honest one available.

How to

1. Before choosing an action, ask whether cause and effect are self-evident, expert-knowable, only visible in retrospect, or genuinely unknowable.
2. Name the context out loud to your team so the diagnosis becomes a shared and contestable claim, not a private assumption.
3. Treat 'disorder' as a live possibility whenever people are arguing past each other, and break the problem into parts that each sit in a clearer context.

Watch out for

• Defaulting to the context you are most comfortable operating in rather than the one the evidence supports.
• Assuming a single situation belongs to one context when it is actually a bundle spanning several.

Underneath every situation sits a structural fact that most decisions never examine directly: how tightly, and how visibly, cause is bound to effect. This is not about how hard the problem feels. It is about whether the link between action and outcome can be seen, and whether it stays put once you find it.

The relationship comes in grades. At one end it is clear and stable, so evident that the same action produces the same result and everyone can agree on why. A step in from there, the link is stable but not obvious, knowable only to someone with the training to trace it, where more than one defensible answer may exist. Further along, cause and effect can be understood only in hindsight; the pattern is real but appears after the outcome, not before the choice. At the far end, there is no discernible link at all, and searching for one wastes time you may not have.

What matters for a leader is that this structural condition is prior to everything else. It does not depend on preference or skill. It governs what kind of knowing is even available, and therefore what kind of response can work. Read the causal structure wrongly and the diagnosis that follows will be wrong too, no matter how carefully the rest of the reasoning proceeds. The relationship between cause and effect is the ground the whole decision stands on, and it changes shape more often than leaders expect.

How to

1. For a given situation, test whether an expert could reliably predict the outcome of an action in advance — if not, you are past complicated into complex.
2. Ask whether the relationship between what you do and what results is repeatable, or whether it emerges differently each time.
3. Distinguish 'we don't yet know' (complicated, knowable with analysis) from 'it cannot be known ahead of time' (complex or chaotic).

Watch out for

• Confusing your own ignorance with genuine unknowability, which pushes you to abandon analysis prematurely.
• Assuming causality is stable when you are in a system that is actively reorganizing itself.

Each context calls for a different sequence of moves, and the sequences are not interchangeable. In a simple context, where the causal link is plain, the leader senses what is happening, categorizes it against known patterns, and responds with the established practice. In a complicated context, where expertise is required, the middle step changes: sense, then analyze, then respond, because the right answer has to be worked out rather than recognized.

Complexity breaks that logic. When cause and effect are visible only in retrospect, there is no pattern to categorize against and no analysis that will reveal the answer in advance. The move becomes probe, sense, respond: run a safe experiment, watch what emerges, then amplify what works and dampen what does not. Chaos inverts the order again. When nothing is stable, the leader acts first to establish some footing, then senses whether the situation is stabilizing, then responds to move it toward a more manageable state. Act-sense-respond trades diagnosis for motion, because in chaos motion is what buys the room to diagnose.

The discipline is matching the method to the diagnosis rather than importing a favorite from another domain. Sense-analyze-respond is sound leadership in a complicated world and a trap in a complex one. The fit between context and method is what produces good decisions, and a leader's ability to switch methods as the context shifts is what keeps that fit from decaying.

How to

1. In simple contexts, categorize the situation against known best practice and respond directly — do not overthink it.
2. In complicated contexts, bring in experts to analyze, but keep decision authority with the leader who weighs options.
3. In complex contexts, run small probes to see what patterns emerge, then amplify what works and dampen what doesn't.
4. In chaotic contexts, act decisively to establish order first, then sense where stability is forming and respond to move the situation toward complexity.

Watch out for

• Reaching for expert analysis in a complex context, where no expert can predict which intervention will work.
• Waiting to understand before acting in a chaos, where the absence of any action is itself the danger.

Contexts change, and they rarely change politely. A situation that was complicated yesterday can tip into complexity overnight, and a stable domain can fracture into chaos without warning. More awkwardly, several contexts often coexist inside the same problem at the same time, each demanding a different response. A leader locked into one style will handle only the part of the problem that happens to suit that style.

Adaptive flexibility is the capacity to move between methods as the ground moves. It rests on two things: an openness to changing course, which is temperamental as much as technical, and the practical skill of shifting decision style on purpose rather than by accident. Both are needed. Openness without skill produces flailing; skill without openness produces a leader who can, in principle, switch but never does, because switching feels like admitting the first read was wrong.

The default is the enemy here. Every leader has a preferred style, usually the one that earned them their position, and under pressure the preferred style asserts itself hardest, exactly when a different one may be required. Flexibility does not replace competence in any single method; it governs whether that competence gets applied where it belongs. It is what lets a correct diagnosis become a correct response instead of curdling into the same move applied to every context in turn.

How to

1. Identify your default style and the context it fits, then treat every other context as one where you must consciously override that default.
2. Build a personal signal for context transitions — for example, a stabilizing crisis — and rehearse the shift before you need it.
3. Surround yourself with people who operate naturally in the contexts you find hardest, and cede the method to them there.

Watch out for

• Riding a directive crisis style past the point where the crisis has passed, suffocating the emergence you now need.
• Mistaking style-switching for indecision and reverting to your comfort mode under pressure.

Each context carries its own characteristic way of failing, and the failures are seductive precisely because they look like strengths pushed slightly too far. In the simple domain, where established practice reliably works, the danger is complacency and oversimplification: the leader assumes the situation is simple because it has always been simple, and misses the moment it stops being so. Confidence built on repeated success becomes the blind spot.

The complicated domain traps experts. Deep knowledge makes people effective and also entrained, dismissing a novel solution because it does not match what has worked before. The mirror image is analysis paralysis, where the search for the best possible answer consumes the time in which any answer would have helped. Both come from the same source: expertise that has stopped listening to the specifics of the case in front of it.

Complexity tempts leaders toward overcontrol. Faced with a situation whose patterns emerge only after the fact, the instinct is to grip harder, impose structure, demand plans, when the situation actually requires letting patterns surface through safe experiments. The grip strangles the very emergence it depends on. Chaos, most dangerous of all, isolates the leader. In the rush to act, a leader can cut themselves off from dissent, and success in a crisis can breed something close to cult adoration, where followers stop questioning and the leader stops hearing anything but agreement.

A correct diagnosis and a well-chosen method still fail if the leader walks into the trap that comes with the territory. Knowing the trap is native to the context is the first defense against it.

How to

1. In simple contexts, guard against complacency by inviting dissent and staying alert to conditions that would push the situation into chaos.
2. In complicated contexts, deliberately solicit novel ideas from non-experts to counter entrained expert thinking, and set analysis deadlines to prevent paralysis.
3. In complex contexts, resist the urge to impose order and instead tolerate the discomfort of letting solutions emerge.
4. In chaotic contexts, actively seek dissenting voices and create channels that penetrate any bubble forming around you.

Watch out for

• Reading a simple context as permanently stable, which sets up the cliff-edge fall into chaos.
• Letting a crisis elevate you into an unquestioned hero whose directives no one dares challenge.