Better Business Decisions Require Better Information — Part 2

Why and How We Become Confidently Wrong

2026

Better Business Decisions Require Better Information

Part 2 — Why and How We Become Confidently Wrong

Roger E. Henley II · 2026

Purpose

This paper is intended to improve business decision-making under uncertainty in complex operations. It is not an attempt to govern personal beliefs, private reflection, or individual life choices. Its focus is narrower and more practical: how people inside organizations interpret information, form explanations, accept risk, and act inside complex systems where the cost of error can be high and the consequences may appear long after the original decision was made.

The purpose of this paper is to explain why better information alone is not enough. Even when relevant data exists, business decisions can still degrade when people search for confirming evidence, defend early framing, inherit untested assumptions, mistake confidence for correctness, or treat agreement as proof. These are not unusual flaws belonging to a few people. They are common human tendencies that become more dangerous under operational pressure, incomplete visibility, time constraints, and organizational friction. In business settings, those tendencies can distort root-cause analysis, weaken escalation, hide residual risk, and turn system effects into people blame.

This paper is therefore meant to provide an operational bridge between decision quality and a broader risk-and-opportunity discipline. A mature risk and opportunity process should do more than satisfy an audit requirement. It should force clearer hypotheses, visible assumptions, explicit ownership, evidence of control effectiveness, defined reassessment triggers, and a more disciplined search for refutation before action hardens. In that sense, this paper is not mainly about psychology. It is about building a repeatable business framework that makes human judgment more reliable when the environment is complex and the cost of being wrong is real.

This purpose also aligns with the intent of ISO-based management systems. Those systems require more than documentation; they require disciplined evaluation of risk, control effectiveness, change, escalation, and review. They are most valuable when they help the business challenge assumptions early, surface uncertainty before it compounds, and make decisions in a way that is evidence-based, reviewable, and capable of improvement over time. The goal here is to support that discipline in practice, not merely in procedure.

This paper is the second part of a five-part series. Part 1 addressed how incomplete, fragmented, or poorly structured information leads to weak business decisions. Part 2 addresses the next layer: why capable people can still become confidently wrong, and why structured challenge, risk framing, and evidence-based review are necessary if the business wants decisions to survive contact with reality. Part 3 addresses what happens to a correct analysis in the exchange that is supposed to turn it into a decision, Part 4 specifies the technical review as a control, and Part 5 specifies the information infrastructure that control depends on. Together the five parts describe a single discipline — surfacing weak signals early and testing conclusions before action hardens, so problems are caught while they are still inexpensive to correct.

Introduction

Part 1 of this series argued that business decisions go wrong when the structure of a problem is hidden by incomplete, fragmented, or badly framed information. That was the information problem. This paper addresses the next constraint: the human problem.

Business decisions are not the same as personal decisions. In personal life, many judgments can be local, reversible, and absorbed privately. In business, decisions often carry delayed consequences, cross-functional effects, formal accountability, and non-trivial cost of error. That means business judgment needs more structure than intuition alone can provide.

Even when better information exists, we do not process it neutrally. We distort how we search, interpret, and defend conclusions. We inherit assumptions without noticing them. We mistake confidence for correctness. We treat agreement as reassurance. We stop searching once a story feels good enough.

These are not failure modes of other people. They are ordinary human tendencies. All of us are vulnerable to them, especially under pressure, time constraints, incomplete information, and organizational friction. The goal is not to criticize people for being human. The goal is to understand how intelligent, experienced people become confidently wrong, and how a disciplined risk-and-opportunity framework can reduce that failure mode.

Good business decisions do not come from louder conviction. They come from better structure, better evidence, and a better process for challenging the current view before action hardens.

The Hidden Truth

The human mind is not optimized for truth-seeking. It is optimized for speed, coherence, social cohesion, and survival.

In day-to-day life, that is often useful. People cannot stop to perform a full causal analysis every time they interpret a situation. They rely on pattern recognition, prior experience, intuition, and compressed models of reality. That is efficient. It is also dangerous for business decision-making, where the environment is complex, multi-variable, and only partially observed.

In manufacturing — across engineering, maintenance, quality, operations, technology, and leadership — the most important decisions rarely occur in simple environments. They happen inside systems with hidden dependencies, delayed effects, cross-functional interactions, measurement limitations, and incomplete visibility. In those conditions, the human tendency to prefer a fast, coherent story over an uncertain, evidence-heavy one becomes a liability. In business decision-making, context governs meaning.

The real issue is not that we lack intelligence. The issue is that intelligence can make the problem worse when it operates inside a weak reference frame. Skilled people can build persuasive justifications for weak conclusions. They can explain away exceptions, reinterpret conflicting data, and defend assumptions long after those assumptions should have been challenged. The smarter the person, the more elegant the rationalization can become.

We all interpret findings through a model of the world. All models are wrong in the strict sense, but some are useful. Recognizing that helps us frame conclusions with more discipline and less arrogance.

This is why confident error is common in organizations. The wrong answer often arrives wearing the clothes of competence and consensus.

Confirmation Bias: Searching in the Wrong Direction

One of the most common and most damaging patterns in decision-making is confirmation bias.

People tend to agree more readily on things they can directly experience: a valve is leaking, a tank is empty, a motor is hot. They diverge much more on things they cannot directly observe: statistical causality, long-horizon quality effects, future risk, control effectiveness, or the economic consequence of a decision made today. In those domains, the same report can support sharply different conclusions because interpretation depends more heavily on priors, models, proxies, and assumptions.

Once a person forms an initial hypothesis, they tend to look for evidence that supports it rather than evidence that could disprove it. This happens so naturally that it often feels like ordinary investigation. It is not. It is directional search, and it reinforces the tendency to preserve the current idea instead of testing it.

That distinction matters.

If a team already believes a capital project is necessary, it will begin noticing every downtime event, every operator complaint, and every maintenance burden that supports urgency, while underweighting normalization, alternative fixes, and the actual cost of error. If another group already distrusts the same project, it will begin noticing every budget pressure, implementation risk, and governance burden that supports delay, while underweighting lost opportunity. The data set can be similar. The direction of search changes.

That tendency is costly. It slows learning, weakens correction, and makes innovation more difficult than it needs to be.

Supporting evidence feels efficient because it creates momentum. It reduces uncertainty. It gives people a reason to act. It protects time already spent. It protects political position. It protects ego. It protects prior statements already made in front of other people.

Disconfirming evidence does the opposite. It forces rework. It introduces delay. It reopens questions people thought were settled. It threatens ownership, status, and narrative stability. That is why organizations tend to underinvest in it.

But that is precisely where decision quality lives.

Key point: Supportive evidence is easy to find once a hypothesis exists. The real test of a decision is whether the current explanation survives honest contact with the strongest evidence against it.

A mature decision process therefore cannot stop at the question:

What evidence supports this view?

It must also ask:

What evidence would prove this view wrong?

That is not a philosophical luxury. It is an operating requirement.

Framing: The World Model You Did Not Notice You Were Using

People do not interpret facts in a vacuum. They interpret them through a world model or reference frame.

A world model is the set of assumptions, experience, incentives, language, and patterns through which a person understands reality. It determines what feels important, what seems plausible, what gets ignored, and what kinds of explanations feel natural.

That means the same data can produce different conclusions depending on who is looking at it.

Operations may see an execution problem.
Quality may see a control problem.
Maintenance may see a reliability problem.
Engineering may see a design or process interaction problem.
Technology may see a visibility, instrumentation, or systems integration problem.
Finance may see a cost and variance problem.

Each of those views may contain part of the truth. None of them is automatically the whole truth.

This is where framing becomes dangerous. Once a problem is framed a certain way, everything downstream begins to organize around that framing. Meetings shift. Metrics shift. Escalations shift. Corrective actions shift. People begin defending solutions to the framed version of the problem rather than questioning whether the frame itself is incomplete.

The distillery example from Part 1 makes this clear.

If a dashboard shows lower yield concentrated on second shift, the fast story is that second shift is underperforming. That feels reasonable. It creates an accountable target. It suggests obvious countermeasures: tighter supervision, stronger instructions, more oversight, additional checks.

But if second shift disproportionately inherits difficult startup conditions, softer utility performance, and a more difficult material lot mix, then the visible pattern is real while the interpretation is wrong. The shift is not the cause. It is the location where the hidden interaction becomes visible.

This is the failure mode that turns system behavior into people blame.

Key point: A framed problem is not necessarily a false problem. It is an incomplete one. The danger is that once it feels coherent, people stop searching for the missing structure.

Assumptions Kill Context

Bad decisions are rarely built on facts alone. They are built on facts plus assumptions, and the assumptions are often doing most of the work.

We can never have all the information. Complex processes do not permit perfect knowledge. That is not the problem. The problem is pretending assumptions are certainties instead of treating them as uncertainties. When assumptions are explicit, they can be challenged, assigned likelihoods, and managed as risks. When they remain implicit, they quietly control the decision.

Assumptions are compressions of uncertainty. They are substitutes for missing context. They allow people to proceed without resolving everything. That is sometimes necessary. It becomes dangerous when assumptions harden into invisible premises.

Examples are everywhere:

The alarm would have caught it.
If the test is accurate, the result is probably real.
If the aggregate metric moved, the process must have changed.
If one shift looks worse, that shift is probably the issue.
If nobody escalated it, it must not be serious.
If multiple people agree, the interpretation is likely correct.
If a control exists, it must be effective.

Each of those statements may be true in a given case. None of them is safe to assume by default.

Untested assumptions are dangerous because they erase context.

They hide base rates.
They hide selection effects.
They hide conditioning variables.
They hide the difference between detection and prevention.
They hide the difference between correlation and cause.
They hide the fact that multiple subsystems may be interacting at the same time.

Key point: Once context is lost, confidence rises faster than correctness. This is one of the most important asymmetries in decision-making: confidence can increase while truth quality decreases. A cleaner story often feels better than a truer one.

The practical implication is blunt:

Every important decision should force its assumptions into the open.

If the assumptions are not written down, challenged, and paired with evidence, they will quietly drive the decision anyway.

Why We Resist Disconfirming Evidence

People often speak as if the only barrier to good decisions is lack of information. In practice, there is another barrier: most of us do not like information that destabilizes our current view.

That resistance is not always conscious. It can come from several places at once:

Identity protection: If a person has publicly backed a position, contradictory evidence can feel like a threat to competence rather than a normal part of learning.
Effort protection: If a team has already invested time, money, or political capital into a line of action, changing direction can feel like admitting waste.
Social protection: People are more comfortable agreeing with a strong narrative than reopening uncertainty in front of peers or leadership.
Narrative closure: Organizations like clean explanations. Ambiguous answers are harder to communicate, harder to govern, and harder to act on.
Accountability avoidance: A simple local explanation often creates a simpler accountability path than a complicated cross-functional explanation.

These pressures do not disappear just because the organization says it values data. In many cases, they become stronger precisely because the data is being interpreted inside a social and political system.

That is why the search for refutation must be procedural, not optional.

This is one reason formal disciplines such as documented risk registers, Management of Change, corrective action, and management review are useful. They create a repeatable place to record assumptions, evidence, owners, review triggers, and residual risk rather than leaving them to memory, hierarchy, or influence.

If searching for contradictory evidence depends on personal courage alone, it will be inconsistent. It must be built into how decisions are framed, reviewed, and escalated.

Confidence Is Not Evidence

Organizations routinely overvalue confidence.

Confident speech sounds like leadership. Fast answers sound like competence. Clear explanations sound like control. Uncertainty can sound weak, even when uncertainty is the only honest description of the situation.

This creates a serious problem. People begin to confuse a confident messenger with a validated conclusion.

That assumption is dangerous in every function:

A confident vendor claim is not proof of model performance.
A confident root-cause explanation is not a causal demonstration.
A confident operating judgment is not a substitute for context.
A confident engineering recommendation is not a finished risk analysis.
A confident consensus in a meeting is not a controlled experiment.

Key point: Confidence does not equal evidence.

Confidence is a feeling. Evidence is a structure. The two are not equivalent.

An organization that rewards certainty and consensus more than rigor will systematically bias itself toward premature closure. It will encourage people to defend answers before those answers are strong enough to deserve defense.

The correction is not to eliminate judgment. That would be unrealistic and undesirable. The correction is to create a culture — and then a framework — in which judgment is expected to survive challenge.

The standard should not be:

How strongly can this be argued?

The standard should be:

How well does this explanation survive contact with contradictory evidence, alternative hypotheses, and missing-context review?

The Majority-Must-Be-Right Fallacy

One of the most common shortcuts in decision-making is to treat agreement as validation.

If many people believe something, others begin to assume it is probably true. The belief feels safer. It plays into our natural tendency toward social cohesion, carries less personal risk, offers the comfort of social cover, and reduces friction.

But truth is completely independent of the number of people who believe it.

Reality does not count votes.

A majority can be wrong. A leadership team can be wrong. A technical group can be wrong. A whole company can be wrong. An industry can be wrong.

Consensus can be useful as a coordination mechanism. It can help organizations move. It can signal that people share assumptions, vocabulary, or priorities. But it is not a validation method.

In fact, consensus often becomes most dangerous when it arrives too early. Once a majority opinion forms, minority interpretations face more resistance. People stop looking for disconfirming evidence because the social need appears satisfied. The group shifts from investigation to reinforcement.

Key point: This is how weak explanations become embedded practice.

The majority-must-be-right fallacy is especially dangerous when combined with incomplete data. A group can converge on the same wrong answer for the same wrong reasons while feeling increasingly secure because others agree.

The operational discipline required here is straightforward:

Do not ask only whether there is agreement.
Ask whether the agreement was stress-tested.
Ask what contrary evidence was considered.
Ask what assumptions are shared across the group.
Ask what nobody in the room is incentivized to challenge.

Alignment is useful. It is not proof.

The Organizational Cost of Defending the First Story

When organizations defend the first coherent explanation instead of testing it, they pay twice.

They pay once for the original issue.

Then they pay again for the cost of the wrong response.

That second cost is often larger than people realize.

It shows up as:

wasted troubleshooting effort
unnecessary holds and delays
bad capital allocation
extra material or chemical usage
misdirected maintenance work
repeated investigations into symptoms rather than causes
avoidable supervision burden
trust damage between departments
blame on people for system effects
erosion of signal trust in data and reporting

Once that cycle repeats enough times, a deeper problem forms: the organization becomes less willing to believe its own information. People start relying more heavily on tribal knowledge, politics, and personality because the formal system has not consistently delivered correct conclusions.

At that point, the enterprise is no longer merely making isolated bad decisions. It is degrading the conditions required for good decisions in the future.

That is why decision quality is not a soft topic. It is a necessary operational capability.

Why Risk and Opportunity Framing Helps

A useful decision process does not ask only whether a current explanation feels convincing. It asks what is at stake if that explanation is wrong, what value exists if it is right, and what evidence most efficiently reduces uncertainty.

That is why risk and opportunity framing matters.

A proper risk-and-opportunity approach forces a decision out of the world of narrative preference and into the world of consequence, uncertainty, controls, alternatives, ownership, and timing.

That change in structure matters because it pushes us to ask harder questions:

What is the current hypothesis?
What observations support it?
What observations challenge it?
What assumptions are embedded inside it?
What alternative explanation could also fit the same facts?
What is the risk if we act now and this is wrong?
What is the risk if we delay and this is right?
What is the opportunity if this interpretation proves correct?
What controls exist already?
What evidence shows those controls are actually effective?
What condition should trigger reassessment?
Who owns the decision, the follow-up, and the residual risk?

That is a far better operating posture than simply asking whether a conclusion feels reasonable.

Risk and opportunity framing does not remove uncertainty. It disciplines it.

The Required Shift: From Advocacy to Structured Challenge

Most meetings are structured as advocacy exercises. People bring their interpretation and try to persuade others.

That format is insufficient.

A stronger format is structured challenge, ideally supported by a standardized risk-and-opportunity assessment.

In structured challenge, the goal is not to win the argument for the current hypothesis. The goal is to expose the current hypothesis to credible challenge early enough to improve the decision before the cost of error compounds. That requires a few shifts:

Separate the hypothesis from the person. An idea being wrong is not a character failure. It is normal. If people feel personally attached to being right, they will hide uncertainty and resist correction.
Reward useful contradiction. The person who surfaces the fact that changes the decision is helping the business, not slowing it down.
Require alternative explanations. No important decision should proceed with only one causal story on the table when multiple are plausible.
Make uncertainty explicit. Uncertainty should be named, not buried. Hidden uncertainty causes false precision.
Force time-bound reassessment. Every major decision should include a condition or date on which the current view must be revisited.

These are not academic habits. They are practical safeguards against institutional self-deception.

A Practical Operating Standard for Better Decisions

For important decisions, investigations, escalations, and risk reviews, the organization should require a standard decision frame. A one-page version is enough if it is done honestly.

Minimum decision frame

I. Decision or issue under review: Define the scope clearly.
II. Current working hypothesis: State the present explanation in plain language.
III. Evidence supporting the hypothesis: List the direct observations that point in its favor.
IV. Evidence challenging the hypothesis: List the strongest facts, anomalies, or counterexamples that do not fit.
V. Alternative hypotheses: Provide at least two other plausible explanations if the issue is material.
VI. Key assumptions: Write down the assumptions being treated as true without full proof.
VII. Missing information: Identify the data or context that would most meaningfully reduce uncertainty.
VIII. Risk if the current hypothesis is wrong: Quantify or describe the downside of acting on a false view.
IX. Opportunity if the current hypothesis is right: State the upside of correct action or correct understanding.
X. Existing controls: List the controls that should already detect, reduce, or contain the issue.
XI. Evidence of control effectiveness: Do not assume controls work because they exist. State what evidence shows they are effective.
XII. Trigger for reassessment: Define what event, metric, threshold, or date would force re-evaluation.
XIII. Owner: Identify who is accountable for follow-up, escalation, and closure.

This structure does three things at once:

It exposes assumptions.
It forces disconfirming evidence into view.
It ties interpretation to consequence and ownership.

That is what better decision discipline looks like in practice.

What Leaders Need to Do Differently

Leaders shape decision quality less by providing answers and more by shaping the rules under which answers are allowed to harden.

That means leaders should do the following consistently.

Ask for the refutation case

When someone presents a recommendation, ask: What is the strongest evidence that this is wrong?

Refuse unsupported certainty

If the claim is stronger than the evidence, push it back to the level the evidence can support.

Separate urgency from haste

Fast action is sometimes necessary. Premature certainty is not.

Prevent people-blame shortcuts

When a human explanation appears first, check whether a systems explanation was actually ruled out or merely ignored.

Require context before comparison

Normalize for mix, conditions, operating mode, and inherited constraints before drawing comparative conclusions.

Reward disciplined escalation

Escalation should not be treated as weakness. It is often the act that prevents local misinterpretation from becoming enterprise cost.

Enforce evidence standards for residual risk

No one should be allowed to claim a control is effective or a residual risk is acceptable without objective support.

If leadership does not do these things, the organization will drift toward fast narratives, shallow consensus, and recurring misdiagnosis.

What This Means in Practice

The first step is recognizing that better decisions require better information. That principle is foundational.

The next step is to recognize that better information alone is not enough if people are allowed to search, frame, and defend conclusions in undisciplined ways.

A strong risk-and-opportunity practice can become the operating mechanism that closes that gap.

Used correctly, it can do more than satisfy an audit requirement. It can become the structure that forces clearer ownership, better assumptions, better evidence, better escalation, better reassessment, clearer authority for accepting residual risk, and better definition of what "effective control" actually means in practice. It can reduce the frequency with which the business mistakes a partial story for a complete one. It can reduce the tendency to blame the visible location of a problem rather than identifying the actual system interaction producing it. It can improve how decisions are made across engineering, technology, operations, quality, maintenance, and leadership.

But that only happens if the process is treated as decision-critical rather than ceremonial.

If it is treated primarily as a documentation exercise, it will not improve decisions.
If it is treated primarily as a color-matrix exercise, it will not improve decisions.
If it is used mainly to justify the current story rather than challenge it, it will not improve decisions.

The process has to help the business think better, not simply record what people already think.

Closing Thought

Most wrong decisions do not begin with bad intent. They begin with an incomplete view, a coherent story, and our tendency to stop searching once a story feels "good enough."

That is why we become confidently wrong.

We do not merely lack information. We have a strong tendency to protect the first explanation that appears to fit. We search for support when we should be searching for refutation. We inherit assumptions without exposing them. We take agreement as reassurance. We confuse conviction with proof.

The answer is not paralysis. The answer is disciplined judgment, governed by a standardized framework.

Disciplined judgment means asking better questions before action hardens:

What are we assuming?
What evidence are we missing?
What would prove this wrong?
What alternative explanation also fits?
What is the risk if we act on a false story?
What is the opportunity if we are right?
Who owns the next review?

That is the real shift. The goal is not to remove human judgment. The goal is to make human judgment strong enough to survive contact with reality.

That is what a mature risk-and-opportunity culture should do. That is what better decision-making requires.

Roger E. Henley II — Part 2 of a five-part series on decision quality in complex operations. Preceded by Part 1: "Why Acting on Partial Data Feels Right and Often Is Not." Continued in Part 3: "When Being Right Is Not Enough," Part 4: "The Technical Review Standard," and Part 5: "The Decision Infrastructure Standard." github.com/reh3376 · linkedin.com/in/rogerehenley

More writing →