The Leadership Friction  |  Edition No. 30  |  September 22, 2026

The Human Challenge in an AI World  —  Part 6 of 8

Not Everything Important Shows Up in the Data

The data can look clean while something underneath it is already off. This edition is about the signals that never make it into a report, and why the best leaders learn to read past the dashboard.

The dashboard can be clean, and the organization can still be in trouble.

The survey results come back fine. Meeting notes show agreement across the table. The numbers are trending in the right direction.

And something still feels off.

Maybe turnover has crept up a little. Maybe the room has gotten quieter than it used to be. Maybe decisions that took a day now take three, and nobody can quite explain why. Maybe the real conversation is happening in the hallway after the meeting ends, not during it.

Here’s what I’ve learned after years of sitting across the table from leadership teams: data can tell you a great deal. It cannot always tell you what is happening underneath it. Leaders who confuse “the numbers look fine” with “everything is fine” are usually the ones who get surprised six months later.

What AI Can Actually See

AI is genuinely good at working with what is available. Feed it survey responses, meeting transcripts, engagement scores, performance data, and written feedback, and it will find patterns a person could miss for weeks.

Here’s the limit nobody talks about enough. If people withhold information, avoid conflict, or carefully manage what they say, the data is incomplete before AI ever touches it. That’s not a flaw in the tool. It’s a leadership reality that existed long before AI arrived, and AI simply inherits it along with everything else.

Silence Says Something Too

Some of the strongest signals in an organization never make it into a dashboard because nobody puts them into words in the room where it would matter.

A person who used to ask sharp questions in meetings stops asking them. A leader who used to push back on the CEO starts going along with everything. People agree fast in the room and resist everything afterward. An employee gives the right answer on a survey, then does something else entirely once they’re back at their desk.

None of that shows up cleanly in a report. All of it is information, and it’s often the most important.

Gallup has found that only 3 in 10 U.S. employees strongly agree that their opinions seem to count at work. That means leaders should be careful about assuming that what they hear in meetings, surveys, or formal feedback channels represents everything employees are thinking. Sometimes the missing information is not absent because there is nothing to say. It is absent because people are not convinced it is worth saying.

Trust Shapes What the Data Is Worth

Leaders often treat survey results and sentiment tools as a direct window into what employees actually think. It isn’t that simple.

The accuracy of what you get back depends heavily on how safe people feel being honest. If someone doesn’t believe candor is safe, what you’re reading is what they were willing to say, not necessarily what they believe.

Bad data isn’t always inaccurate data. Sometimes it’s incomplete data, produced by an environment where people have decided the whole truth isn’t worth the risk.

Relationships Carry What Numbers Cannot

Leaders who know their people well tend to notice things before those things become measurable. They pick up on a change in tone before anyone mentions it, on hesitation that wasn’t there last month, on tension between two leaders who used to work well together and don’t anymore. They notice the frustration sitting underneath an agreement that came too easily, or the moment someone starts checking out mentally before they check out for good. Sometimes they notice that a decision was technically accepted in the room but never actually picked up by the team.

That kind of read comes from proximity, not from a report. AI can help surface patterns across a large amount of information. It cannot replace the context a leader builds by paying attention to real people over real time.

History Explains What the Numbers Cannot

A new process can look entirely reasonable on paper. But if the organization handled the last change badly, the team isn’t reacting to this change. They’re reacting to that one.

A new leader might assume a tense conversation is about the current initiative, when the room is actually responding to something that happened two years before that leader arrived. The numbers will show resistance. The history is what explains it.

That’s why organizational information rarely holds up when read in isolation. Context isn’t optional. It’s half the story, and it’s the half that doesn’t show up in the report.

Let AI Raise Questions, Not Settle Them

None of this means AI has no place here. Used well, it can help a leader see themes across more information than a person could sort through by hand: which challenges keep surfacing across channels, where sentiment is shifting, which teams seem to be carrying more friction than others.

The value is in what those findings lead to. A good pattern from AI should open a better conversation. It shouldn’t end one. The moment a leader treats an AI summary as the final word instead of the first question, the tool has taken over a job it was never suited for.

I have created AI-supported tools that I use when working with organizations. I take the data from assessments and surveys and look for patterns, themes, and areas of friction.

But I do not stop with the data.

I then interview key people in the organization to understand how what the data is showing lines up with what people are actually experiencing. Sometimes the two reinforce each other. Sometimes the conversations reveal context the data could never have captured on its own.

The Friction Point

The real risk isn’t that AI misses something. It’s that leaders stop looking once the report looks fine.

When the numbers are steady, it’s tempting to treat that as permission to stop asking harder questions. That’s where the drag builds, and it builds well before anyone can point to it: the turnover nobody predicted, the team that agreed to the plan and never really executed it, the resistance that had a two-year history nobody accounted for. None of it shows up as an alert. It shows up months later as a result, and by then it’s expensive to trace back.

Two Moves This Week

What is happening in your organization right now that you know because you’re paying attention, and not because a dashboard told you?