Development has to begin before there is a track record, because by the time somebody has one they have already been developed by whoever gave them the chance.
Which means identifying capability from evidence available in ordinary work, and most organizations are using proxies that measure something else.
The proxies that mislead
Confidence. Measures temperament and prior treatment. It is frequently inversely related to calibration, and it is the single most over-weighted signal in most organizations because it is the most visible.
Articulacy. Somebody who explains things well is easier to assess, which gets mistaken for their being better. The quiet person doing excellent work is harder to evaluate, and difficulty of evaluation gets read as absence of evidence.
Credentials. Establish that somebody completed a programme some years ago. They say very little about judgment, and in fields where the credential is common they say nothing about relative capability at all.
Visibility. Proximity to senior people, which is largely a function of role and location. The person on the night shift or at the second site is not less capable; they are unobserved.
Willingness. Enthusiasm for advancement is not evidence of aptitude for it, and the correlation is weak enough to be worth ignoring.
Seven signals that precede a record
All of these are observable in ordinary work by anybody paying attention.
They ask about the reason rather than the instruction. Somebody who wants to know why the process exists is building a model rather than following a rule, and the model is what handles the case nobody anticipated.
They bring problems early, including their own. The willingness to report something before it is contained is both a character signal and evidence that they can distinguish a developing problem from an ordinary difficulty.
They can state the case against their own view. Ask what would change their mind. A specific answer indicates they have tested the position; an inability to produce one indicates they have adopted it.
They finish things nobody is watching. Follow-through on unsupervised work is difficult to fake and is one of the strongest available predictors.
Other people check things with them. Informal authority is assigned by peers on the basis of accumulated observation, and peers are considerably better positioned than management to make that assessment.
They handle being wrong without either collapsing or defending. How somebody absorbs a correction predicts their trajectory more reliably than how often they need one.
They notice second-order effects. Somebody who says this will work and it will create a problem for scheduling in March is thinking about a system rather than a task.
The people who go unnoticed
Capability is distributed more evenly than opportunity, and certain profiles are systematically missed for reasons that have nothing to do with ability.
People whose work prevents problems rather than solving them, whose contribution is invisible precisely because it succeeded. People in functions that never present to leadership. People who joined without the usual credential and are therefore not on any list. People who are not comfortable advocating for themselves, in a culture that rewards the ones who are.
The correction is a specific question put to supervisors: who in your area would surprise me? That phrasing works better than asking who is high-potential, because it asks about the gap between somebody’s reputation and their actual capability, which is exactly the population being missed.
The only reliable test
Signals narrow the field. They do not settle it, and no amount of observation substitutes for the actual test.
Give somebody a real decision with real consequences and watch what they do. Not a project to manage — a decision to make, where being wrong will cost something and they will own the result.
What you are watching is not whether they get it right. It is how they reason under exposure, what they do when they realise they were wrong, and whether they tell you before you find out.
This is also why identification and development are not separable activities. The test is the development, and the organization learns whether somebody is capable by the same means that makes them so.
Age and tenure are poor filters
Two errors, running in opposite directions, and both are common in institutions that have been stable for a long time.
Assuming somebody in their late twenties cannot hold responsibility, when capability at that age is largely a question of whether anybody has yet given them a chance to accumulate cases.
And assuming somebody with twenty years has judgment, when twenty years of unexamined repetition produces considerably less than five years of decisions with feedback.
The question in both cases is the same and it is about closed repetitions rather than elapsed time: how many decisions has this person made where they found out whether they were right?
Edited by Patrick J. Wolf, PhD