Organization Design

Relative Performance Evaluation and Tournament Pay

Learn why judging people against each other is free noise reduction, exactly how much it is worth in slope terms, and the two conditions under which it stops working and starts destroying the cooperation you were relying on.

  • Advanced
  • 14 min total
  • 14 chapters

What decision this helps you make: Whether to judge a role against an absolute target or against a peer group, how big that peer group has to be before the benefit is real, and whether a promotion contest is the incentive you are actually running.

What this topic is

Relative performance evaluation judges someone against a comparison group rather than against a fixed standard: not "did you hit $1.2M" but "how did you do against the eleven other people carrying similar books this year". Tournament pay is the sharpened version, where the payout depends only on rank and not on the size of the gap — a bonus pool split by position, a promotion, a prize for the best region. Both do the same underlying job. They subtract out whatever happened to everybody, so what remains carries more information about the individual.

Why it matters

Measurement noise is the binding constraint on how strong any incentive can be, and a large share of that noise is usually a shock that hit the whole peer group — a market, a season, a product delay, an economy. Removing it costs nothing and it raises the slope you can afford to run, which is the rarest kind of improvement in compensation design. The catch is equally practical: a tournament pays people for their position relative to their colleagues, which gives them a reason not to help each other, and the help is usually the part of the work nobody measures.

Who should learn it

Anyone running a bonus pool, setting quotas across a team, deciding whether to benchmark a target against the market, or considering a ranked or forced-distribution review system — and anyone who has watched a company run a promotion contest without ever calling it one.

What you will understand

  • Why removing a common shock is free information, and what the informativeness principle says about who belongs in the peer group
  • The arithmetic of peer-group size, and why relative evaluation against one comparator can be worse than no comparison at all
  • What a tournament buys that a linear contract cannot, and the four costs that come with it
  • When heterogeneity kills the incentive entirely, and what handicapping does and does not fix

Prerequisites

Common misconception

"Relative evaluation is fairer, so it must be better." It is not primarily a fairness instrument; it is a noise-reduction instrument, and its value comes from a statistical property rather than a moral one. Two consequences follow that surprise people. Relative evaluation against a small peer group can be worse than absolute evaluation, because the peer average is itself noisy and subtracting it adds that noise back in. And a tournament between people of visibly unequal ability provides almost no incentive to either of them — the strong contestant coasts and the weak one gives up — which is why a ranked system frequently demotivates precisely the people it was designed to sort.