Essays · Leadership · Psychology · AI and work

Every label is a prediction: a leadership lesson about people

Gifted, high potential, not leadership material. Every label is a guess about the future. Here is how leaders keep labels as doors, not walls.

I give tests for a living. Most weeks I still sit across a small table from a child, open a test kit, and start a stopwatch. Afterward I sit across a different table from that child’s parents and try to explain what the numbers mean and what they do not.

I believe in good assessment. I have seen a well-done evaluation open doors for a child that years of good intentions could not. And the most important thing that work has taught me applies just as much in a boardroom as in a school.

Every label is a prediction.

“Gifted” predicts that a child will thrive with advanced work. “High potential” predicts that an employee will succeed in a bigger role. “Not a culture fit” predicts that a candidate will not work out. “Not leadership material” predicts a ceiling.

Some of those predictions are wise. Some are wrong. All of them are guesses about the future, made by particular people with particular purposes, using particular evidence, at a particular moment. The trouble starts when we forget that, and the prediction hardens into an identity.

I have seen it from every side of the table

As a school psychologist, I give the tests whose scores go into a child’s folder. As a principal and superintendent, I signed off on the cutoffs and programs that decide which children get noticed. As a senior manager at Oracle, I saw how large organizations rate and sort their people, and how much rides on those ratings.

The settings were different. The pattern was the same. Once a label is written down, people stop asking where it came from.

Labels do not just describe. They instruct.

Most of us think a label describes what a person already is. But a label also tells people what to do. It tells teachers what to teach, managers what to assign, and the person inside the label what to expect of themselves.

I call this the expectation echo. What we expect of someone shows up in a thousand small decisions: what we hand them to work on, how long we wait for their answer, whether they get the hard problem or the easy one. Those choices shape what they learn and how they see themselves. Then we measure them again, find what we expected, and congratulate ourselves on our accuracy.

I want to be careful here, because the famous story about this is bigger than the evidence. Robert Rosenthal and Lenore Jacobson’s 1968 book, Pygmalion in the Classroom, reported that randomly chosen “bloomers” gained more than their classmates. The effect showed up mostly in the youngest grades, and other researchers questioned the methods. When psychologists Lee Jussim and Kent Harber reviewed decades of research on teacher expectations, they concluded the effects are real but usually small, often fade over time, and may be stronger for students who are already stigmatized or struggling.

So here is how I hold it. One person’s private belief probably does not rewrite anyone’s future. But beliefs rarely stay private. They get built into a schedule, a track, an org chart, a succession plan, a software system. That is where the echo gets loud, because structures do not have good days or bad days. They just keep echoing.

The most common label in business

Take the most common promotion decision there is: who gets to be a manager.

In a study of sales workers at 214 firms, economists Alan Benson, Danielle Li, and Kelly Shue found that companies tended to promote their best salespeople, even when other traits better predicted who would make a good manager. The result was costly mismatches, the old Peter Principle showing up in real data.

“Top performer” is one label. “Ready to lead” is a different label. Most organizations treat the first as proof of the second. The best salesperson becomes a struggling sales manager, and the colleague who was always helping everyone else close deals never gets considered, because her own numbers were merely good.

Label debt

Engineers know about technical debt: shortcuts that work fine today and quietly pile up costs until something breaks. I think leaders carry label debt. It is every rating that was never revisited, every “not ready” that came out of a single meeting, every filter nobody remembers adding.

Label debt does not show up on the balance sheet. It shows up as the talent you lost and the leaders you never developed.

Artificial intelligence raises the stakes. Prediction is almost all AI does, which means every question we should have been asking about human labels, we now have to ask about machine labels too, and faster. A system trained on yesterday’s decisions learns who was allowed to succeed, not who could have. Used with care, the same technology can help pay label debt down. Used carelessly, it runs the debt up at a speed no organization has seen before.

The DOOR Test

The tool I use is four questions. I designed it to fit on an index card, and it works on any label, whether a teacher wrote it, a psychologist measured it, a manager assigned it, or an algorithm produced it.

  • D: Defined by whom? Who decided what this label means, and what counts as evidence?
  • O: Observed how? What evidence produced it, under what conditions, and how sure are we?
  • O: Omitted what? What abilities, experiences, or opportunities did the measure never see?
  • R: Result, door or wall? Does this label open access to help and opportunity, or close it?

A good label is a door. It is specific, temporary, tied to support, and open to revision. A bad label is a wall.

Picture a manager who has quietly filed a soft-spoken analyst under “no executive presence.” Defined by whom? One senior leader’s personal style. Observed how? Two presentations, both to the same tough audience. Omitted what? The fact that she runs the cross-team meeting nobody else will take on. Result? No stretch assignments, no visibility, no path. That label is a wall, and nobody built it on purpose.

Smart is a verb

Potential is not something a test discovers or a machine calculates. It is something people and systems do with a person: notice it, name it carefully, support it, and keep checking.

I wrote Who Gets to Be Smart? for parents, educators, and leaders who want to be the person who looks again. In a school, that might be the adult who sees the child the folder missed. In a company, it is the leader who asks who the system got wrong, not just how often it was right.

The next play

  • Write down one label you use, even privately, about someone on your team. Run the four DOOR questions on it.
  • Find one rating or review that is more than a year old and revisit it with fresh evidence.
  • In your next talent or promotion conversation, separate “top performer” from “ready to lead” and discuss them one at a time.
  • Ask two questions out loud in that meeting: who might we be missing, and who did we get wrong last time?