Almost every brief opens with a number of years, and almost every automated filter enforces it. The research says it is one of the weakest predictors available, which makes it the most expensive habit in recruitment.
"Minimum five years' experience." It appears in most briefs, it is the easiest criterion to enforce automatically, and it is close to the bottom of the evidence table.
Schmidt and Hunter's meta-analysis puts the validity of job experience for predicting later performance at about 0.18, adding roughly 0.03 over general mental ability alone. The relationship is also non-linear: most of the learning happens in the first couple of years, after which additional time adds very little.1
Against that, structured interviews and general mental ability sit around 0.51 in the same analysis.1 Educational credentials add little within a comparable applicant pool.
A five-year minimum is a filter that removes candidates on a criterion with about a third of the predictive power of the interview you were going to run anyway.
Because it is easy, defensible and feels like rigour. A hiring manager who says five years is not making a claim about performance; they are compressing several things they have not separated: someone who has seen a full cycle, who will not need close supervision, who has made the mistakes already.
All of those are reasonable requirements. None of them is measured by counting years, and each can be assessed directly.
Each of those is a structured interview question with a scorable answer, which is precisely the method the research ranks near the top.
This is not only a fairness point, though it is that too: year-count filters disadvantage career changers, people who took time out, and anyone whose experience is real but compressed.
It is a supply argument. A five-year minimum on a scarce skill in a small market removes candidates who would perform, in a country where that skill may only exist in a few hundred people. Replacing the proxy with the requirement it was standing in for widens the pool without lowering the bar.
That is a conversation worth having with a client at brief stage, and it is one of the few places where citing research genuinely helps: you are not asking them to lower standards, you are showing them a better-evidenced way to apply the same one.
Structured questions and consistent scoring, so a shortlist rests on evidence rather than a year count.