What Good Judgment Looks Like

In the Lead magazine seeks to inform future leaders, generate global dialogue on leadership and build a community that sees more effective leadership as a prerequisite to building a better world.

It is published by the Buccino Leadership Institute, developing leaders at Seton Hall for over 30 years. This edition reaffirms Seton Hall’s commitment to fostering innovative, ethical and impactful leadership. Stay ahead of the curve — explore the Fall 2026 issue of In the Lead.

---

Judgment is often defined as “the ability to make a decision, or form an opinion objectively, authoritatively, and wisely, especially in matters affecting action; good sense; discretion.” It’s a word we use constantly in leadership. “She has great judgment.” “That was a lapse in judgment.”

Yet most organizations couldn’t tell you how they’d recognize it if they saw it, let alone how to build more of it. As a management consultant for 35 years, I’ve watched leaders exercise judgment in everything from routine promotion decisions to entering new global markets, and the patterns are more consistent than most people expect.  

The pattern I’ve seen repeated across those decades is that good judgment draws on data and analysis without being ruled by it, stays open to a case that cuts against instinct, and can explain the trade-off it lands on rather than just defend the number. It shows up the same way whether someone is weighing a promotion or a market entry, which is what makes it a discipline you can examine and build, not a talent some people are simply born with.

This question matters differently now. Artificial intelligence has made data, analysis and even recommendations cheaper and faster to produce than at any point in my career. That sounds like good news for decision-making, and often it is, but it also shifts the bottleneck. It used to be that the hard part was getting enough information to make an informed call. Increasingly, the hard part is knowing which of the abundant, AI-generated answers to trust, when to override one, and when a confident-looking output is missing the one thing that actually matters: judgment.

How Do You Know If Judgment Is Good or Bad?

The honest answer: Often you don’t, not right away. Judgment gets evaluated on outcomes, but outcomes are noisy. A good decision can produce a bad result, and a bad one can get lucky. What distinguishes good judgment from bad is visible before the outcome arrives, in the process itself.

Good judgment starts from a wide enough aperture. It draws on data, but doesn’t stop there, incorporating experience and multiple points of view before narrowing to a decision. I saw the opposite play out when we measured promotion decisions almost entirely on hours logged on billable client work. The data was clean and easy to defend, but it didn’t capture whether someone spent that time on mundane tasks or on strategic work that moved a client’s business. A leader exercising good judgment looks past the numbers and asks what it represents; one exercising poor judgment lets the number make the decision because it’s easier to defend later.

Good judgment also means being willing to be uncomfortable. It is being willing to challenge an analysis everyone in the room is nodding along to. I worked with a client optimizing their sales and operations planning process. We built an algorithm to determine the most cost-efficient batch sizes and schedules. The output was clean and well-reasoned. But the heads of sales, production and supply chain still met regularly to review it because the “optimal” answer on paper sometimes would have shorted a strategic client account to save a marginal amount on cost. That’s good judgment: Not ignoring the data but knowing when to override it and being able to explain why.

One useful test, independent of outcome: Did the leaders seek out information that disagreed with their instinct? Did they invite perspectives different from their own? Could they explain, not justify, but genuinely explain the trade-off they were weighing? Poor judgment tends to lean on a single data point, resist dissent or default to whatever’s easiest to defend.

Can You Train for Good Judgment?

Yes, but not the way most companies try to. Judgment isn’t a workshop skill; it’s a craft built through repeated exposure to real decisions, with feedback.

Deliberate exposure to ambiguity. Judgment only develops where the answer isn’t obvious from data alone. Leaders protected from ambiguous calls never get the reps. The consultants and clients I have worked alongside who have the sharpest judgment were given real decision authority early, including room to get it wrong on something that mattered, but not catastrophically.

Feedback loops that connect decisions to outcomes. Most organizations fail here. Decisions get made, time passes, and nobody circles back to ask whether the call was right and why. Without that loop, people accumulate confidence, not learning.

Proximity to people with better judgment than you. This is apprenticeship more than training. Watching how experienced leaders weigh a decision — what questions they ask, what they ignore, where they push back — teaches things no case study can.

A wider library of pattern recognition. Much of what reads as “good instincts” is pattern recognition from having seen many versions of a situation before. It’s trainable through case histories, post-mortems and conversations with people who’ve navigated similar terrain. This is exactly where AI can help — not by deciding, but by surfacing more patterns and precedents faster than any one person could recall, expanding the raw material judgment draws on.

What you can’t train is judgment in a vacuum, disconnected from real stakes. Frameworks alone don’t produce judgment any more than knowing chess rules produces a strong player. It comes from playing the game, repeatedly, with honest feedback.

Why This Matters More in the Age of AI

This has become more urgent, now that AI can produce analysis and recommendations almost instantly.

Abundance of analysis doesn’t equal quality of decision. It shifts where the risk sits. When good analysis was scarce, judgment was tested mostly in gathering the right inputs. Now that AI can generate confident-sounding answers in seconds, the test has moved downstream: Can you tell a genuinely sound recommendation from a polished but flawed one, and do you have the standing to override it when it’s wrong?

This is a harder test, for a specific reason: AI output tends to look uniformly confident regardless of how reliable it is. A spreadsheet missing a key variable usually looks incomplete, but a well-written AI recommendation missing the same variable can look just as authoritative as one that has everything right. The good call and the bad call increasingly arrive looking identical, and only judgment can tell them apart.

There’s a quieter risk too: The more leaders lean on AI to produce answers, the less experience they get building the judgment needed to evaluate those answers well. If AI absorbs more of the ambiguous middle ground, running scenarios, drafting recommendations, pre-narrowing options, then leaders risk losing exactly the practice that builds judgment in the first place. As intelligence gets cheap, judgment is what determines whether that abundance produces better decisions or just faster, more confident-looking mistakes.

What’s Needed From a Personal Perspective

The first habit is intellectual humility. This requires a genuine willingness to be wrong and to look for evidence that you are, rather than evidence that confirms what you already believe. The leaders with the worst judgment I’ve encountered weren’t unintelligent; they were simply overly confident with familiar domains and excessively cautious in unfamiliar domains. Leaders need to be aware of their own blind spots and to be open to other inputs that judgment depend on.

The second is the discipline to slow down on decisions that deserve it. Most judgment failures are failures of speed, not insight.

A third, newly important habit is to treat AI output the way you’d treat a smart but inexperienced analyst’s work. It is useful and fast, but not self-verifying. You should treat an AI recommendation the way you would respond to a junior team member’s recommendation.  You should ask “What information did you not have access to?” and “What would change your answer?”

An inexperienced analyst hasn’t yet built the pattern library that lets a senior leader sense when a confident-looking output is missing something, and AI makes it easy to skip the very struggle that builds this capability. Producing a clean analysis no longer proves an analyst understands the problem; it may only prove they prompted well.

Building that instinct now means running the numbers by hand often enough to notice what a shortcut skipped. It also means treating every AI-assisted answer as a draft to challenge, rather than a finished one to submit. Senior leaders have an outsized role to play here: They should be asking the analyst to explain and defend any recommendation. Later, the senior leader should close the feedback loop by circling back on what actually happened, sharing other insights and explaining why a specific decision was made. Skip that investment, and an inexperienced analyst can become fluent at producing answers without ever building the judgment to know which ones to trust.

What’s Needed From a Systems Perspective

Decision rights that match decision risk. When I worked with a pharmaceutical client evaluating 30-plus countries for global expansion, the data and analysis were extensive. We evaluated population, market size, the regulatory environment and the healthcare infrastructure. But the final call was made by the leadership team, not the analysis, because the organization had built decision rights that put the highest-stakes calls in the hands of the people with the most relevant experience and the most at stake in getting it right.

In addition to the analysis, they also considered their personal experience in the markets and where they had relationships. Organizations that let analysts or models make consequential calls by default, simply because the output looks rigorous, are optimizing for speed and against judgment.

Metrics that don’t accidentally punish good judgment. If your measurement systems reward the easily quantified proxy (e.g., hours logged, cost of goods sold, top-line sales) over the harder-to-measure but more important outcome, you train your best people to stop exercising judgment and start optimizing the metric instead. Systems should be designed to make room for the qualitative override, not penalize it.

Multiple points of view built into the process, not left to chance. The sales and operations planning team’s good collective judgment wasn’t an accident; the organization built a recurring forum where sales, production and supply chain operations were required to weigh in together because no single function’s data told the whole story.

Psychological safety to challenge the data. None of this works if people are afraid to say, “I think the model is wrong here.” Organizations that punish dissent, even subtly, end up with leaders who defer to the analysis because that's the safer call to defend later.

Explicit checkpoints where a human, not a model, signs off. Organizations need to deliberately decide which decisions can be automated and which require a human to review the recommendation and have real authority to overrule it. Leave that line implicit, and it gets drawn by default in favor of speed.

Good judgment isn’t a trait some leaders are born with. It’s personal discipline and organizational design, built and reinforced over time. As AI makes intelligence cheap and abundant, judgment is what determines whether that abundance produces genuinely better decisions or just faster, more confident versions of the same mistakes. The leaders and organizations that take it seriously now, while everyone else assumes the tools will handle it, are the ones who will still be making the right calls when it matters most.

Categories: Business