Promotion mismatch
A Socratic walk-through of promotion mismatch — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why does the best worker on a team so often turn into a poor manager?
A team loses its best engineer and gains a mediocre manager, and everyone involved acted sensibly. The promotion was earned, the decision was defensible, nobody was fooled. Laurence Peter turned this into a joke in 1969 — people rise until they reach a post they are bad at, and there they stay. But a joke that keeps coming true is worth taking apart. What exactly does a promotion decision measure, and what does the new job require?
Reasoning it through
REASONING #Ask first what evidence a firm actually has when it promotes. It has a record of performance in the current role — sales closed, code shipped, cases resolved. That is a rich and honest signal, and it is a signal about one thing: aptitude for the job the person is already doing.
Now ask what the next job requires. Sometimes almost the same skills, one step larger; a senior engineer's work is mostly a harder version of a junior engineer's. But management is a change of kind, not of degree. The task is no longer to do the work but to allocate it, to judge other people's judgement, to sit with a struggling colleague, to say no upward. Whether those abilities travel with the ability to close deals is an empirical question, and there is no reason in principle for the correlation to be high.
Follow that through. If a firm ranks candidates on a measure whose correlation with the thing it actually needs is weak, promoting the top of that ranking is barely better than promoting at random — and it is worse than random in one respect, because it reliably removes the best performer from the post where they were best. The firm pays twice: once in a mediocre manager, once in a vacated star.
Why not simply measure the right thing instead? Two reasons, and they are both real. First, managerial aptitude is close to unobservable before someone manages; you can interview for it, but the record you can audit is the record of the old job. Second — and this one is easy to miss — the promotion is not only a selection device. It is a prize. It is what makes everyone below work hard all year. Promote the second-best performer on a hunch about temperament and the tournament stops being credible, so a firm may knowingly accept a worse manager in order to keep the incentive intact.
There is also a purely statistical contributor, before anything about temperament enters. Observed performance is partly ability and partly luck, and the people at the very top of a ranking have, on average, had more of both. Promote them and the luck does not travel; their performance regresses toward the mean, and the decline looks like incompetence in the new post when part of it was never really there in the old one. Edward Lazear made this argument in 2004, and it means some apparent Peter-principle decline needs no mismatch story at all.
The satisfying part is that this stopped being a matter of opinion. Benson, Li and Merrill examined promotion in sales organisations across a large number of firms and found both halves of the claim in the data: sales performance strongly predicted who got promoted, and it was negatively related to how the promoted person's subordinates then performed. Firms really do promote on the wrong signal, and the wrong signal really does anticorrelate with what the new job needs — in that setting, at least.
The analogy
THE ANALOGY #Imagine a school whose entrance test for each year is simply last year's exam. It works beautifully while the subjects build on one another, because the old test and the new material want the same abilities. Then comes the year arithmetic gives way to essay writing — and the test carries on doing exactly what it was designed to do, faithfully selecting the best arithmeticians for a year of essays.
A school finds out within a term and can move the child, at almost no cost to anyone's dignity. A firm discovers the mismatch only after the appointment, and unwinding it means a demotion — which reads as a public verdict, tends to lose the person entirely, and so mostly does not happen. The stickiness, not the selection error, is what makes the mismatch permanent.
Clarifying the model
THE MODEL #The claim is not that good workers make bad managers. It is narrower and stranger: promoting on current-role performance selects for the wrong trait whenever the two aptitudes are weakly related, and it does so most aggressively exactly where current performance is measured best. A sales force with clean individual numbers is more exposed to this than a team whose output is joint and unattributable, because the crisp measure is the one everybody trusts.
Notice also what the mechanism does not require. No one has to be mistaken about what management involves. The firm can know perfectly well that the star seller may not manage, and still promote her, because refusing would cost more in the incentive it destroys than it saves in the appointment. Some mismatch is not an error being made; it is a price being paid.
The honest counterpoint is that firms are not passive about this, and the joke slightly libels them. Dual career ladders let a specialist advance in pay and standing without acquiring reports — widespread in engineering and research, precisely where the aptitude gap is widest. Trial periods, interim appointments, deputising and rotation all generate evidence about the next role before the appointment is permanent. And where firms take those routes, the pattern weakens. The Peter principle describes a strong default, not an inevitability.
A picture of it
THE PICTURE #How to readThe horizontal axis is what a firm can measure at promotion time; the vertical axis is what the new job actually needs, and is invisible until after the appointment. A promotion rule that reads only the horizontal axis takes everything on the right — which merges the top-right quadrant with the bottom-right one. The bottom right is the mismatch, and the star seller sits deep inside it. Lucky quarter is placed just right of centre and low to make Lazear's point: some of that horizontal position was luck and will not survive the move. Quiet organiser, top left, is the cost nobody counts — the person the rule never reaches.
What became clearer
WHAT CLEARED #A promotion is a prediction dressed as a reward, and the evidence it rests on describes a job the person is about to stop doing. Where the two roles want the same abilities that is harmless; where management arrives as a change of kind, the rule reliably selects on the wrong axis, aided by regression to the mean and held in place by the fact that promotions are far easier to grant than to reverse. What makes it durable is not that firms fail to notice, but that the same promotion is doing a second job — motivating everyone who did not get it.
Where to go next
ONWARD #- Why dual career ladders keep collapsing back into a single one, despite being the obvious fix.
- How tournament theory prices the trade-off between a promotion's incentive value and its match quality.
Key terms
TERMS #| Term | What it means |
|---|---|
| Peter principle | the proposition, from Peter and Hull's 1969 book, that people are promoted until they reach a role they perform badly, and then remain there. |
| Regression to the mean | the tendency of an extreme measured performance, part ability and part luck, to be followed by a less extreme one. |
| Tournament theory | the account of promotion as a contest whose prize motivates all entrants, so its incentive value may outweigh its accuracy as a selection device. |
| Dual career ladder | a structure offering advancement in pay and rank along a specialist track, without requiring a move into management. |
Every term the collection defines is gathered in the glossary.