Selection is a methodical process, and most of its error sits in the specification rather than in the measurement.
Why it matters when the plan changes
Considerable care usually goes into the assessment and very little into what it is assessing against. The requirement is typically written from the previous holder of the role, describing the strategy the company was executing before. Personnel selection is described as a methodical process, but method only compounds whatever question it is aimed at. A rigorous process run against the wrong specification produces a confident answer to a question nobody asked.
The tension is between the person and the arrangement. An assessment describes an individual, but the behavioural data exists to answer whether these people, in these roles, under this plan and this pressure, can execute, and where they will not; the outcome depends on the role, the support, the sequence and the interfaces around them. A finding stated as a property of the person will be acted on as one, which is how a structural problem becomes a personnel decision rather than a redesign of the arrangement.
In practice
A division head is assessed thoroughly before an appointment and rated strong on every dimension in the specification. The specification was written before the operating model changed, and the role now depends on settling conflicts between two product lines, which the previous version of the job never had to do. The assessment measured the old role well.
Evidence
Selection is a methodical process, and the methods used are only as good as what they are applied against.
Personnel selection, Wikipedia (2026)A score predicts a later outcome only for the specific use it was validated against.
Predictive validity, Wikipedia (2026)
What it cannot tell you
Leadership assessment measures how a person's capability and behaviour compare with a stated requirement; it says nothing about whether that requirement is correct. A finding is bound to the role, the plan and the arrangement it was checked against, and predictive validity does not transfer automatically to a different role, structure or strategic period.
Questions
Capability, experience and behaviour against a specification of what the role requires, tested through structured interviews, exercises, references and behavioural instruments. The methods themselves are well developed. The specification they are applied against is the part that most often goes unexamined.
Because it decides what the whole process is measuring. A requirement written from the previous holder describes the job as it was under the previous strategy. Every subsequent step is then rigorous about the wrong question, and the rigour makes the answer more convincing rather than more correct.
It should not. Read properly, an assessment describes fit between a person's patterns and what a specific role under a specific plan demands. Fit is two-sided and situation-bound. Converting it into a ranking discards the half of the information that says what to change.
Anything it was not validated for, and any claim about a person's future beyond the role and horizon in question. A finding is about this role, this plan and this arrangement. Stretched further it becomes a statement about a person's worth, which the evidence cannot support.
By asking what in the arrangement produced it before asking anything about the person. Where the plan demands behaviour the role is not set up to support, the available moves are changing the plan, the role or the support. Changing the person is one option among several, not the first.