Managers in calibration sessions trade support for each other's ratings, protecting their own teams and squeezing rivals, while the cost lands on employees whose managers do not trade.
An example
Every year a grocery chain calibrated its store and pharmacy managers under a forced distribution: ten percent at the top, ten percent at the bottom. Four operations directors had sat on the panel for five years and backed one another's top candidates with little discussion. The fifth member, the head of pharmacy operations, was new and argued each case on the numbers. Over three cycles her managers landed in the bottom tier twice as often as anyone else's, because whenever the curve forced someone down, the other four voted together. One of her pharmacy managers, whose store had the best prescription growth in the chain, was rated below a store manager whose location had lost money. After the third cycle, two of her strongest managers left for a competing pharmacy chain.
Why it works
Trading looks easier: defending your own people alone reads as bias, while defending them inside a standing pact makes the argument look like the room's. Voting on merit costs more, because the lone merit voter has no coalition behind their own people, cannot retaliate against anyone, and becomes the predictable source of the forced-down ratings. Nobody's view of the candidates changes; only the cost of each position does. It needs repeated cycles, a small panel, weak scrutiny of voting patterns, and a forced distribution that makes trades worth making.
The thinking behind it
Robert B. Cialdini Influence (1984) describes reciprocity: people feel obliged to repay what others give them. In repeated calibration cycles, support for a colleague's top candidate creates an obligation to return the favor next time, and the trading sustains itself without any explicit deal. Varying panel composition breaks the repeated relationships that keep those obligations alive.
Oliver E. Williamson Markets and Hierarchies: Analysis and Antitrust Implications (1975) assumes that some people act opportunistically, seeking self-interest with guile, and that organizations exist partly to safeguard transactions against it. A calibration panel is a transaction among managers whose costs fall on absent employees. Outside chairs, varied panels, and evidence the panel cannot control are safeguards of that kind, which is why they work where appeals to merit do not.
Warning signs
Ratings for the same managers' teams rise every cycle while a non-trading or absent manager's teams fall. Discussion of certain candidates is short and unanimous. Forced-distribution cuts land repeatedly on one manager's people. Ratings drift away from measured results in a pattern that follows the manager more than the employee.
How to defend yourself
Exhorting the panel to vote on merit will not change anything; the incentives have to change.
- If you are the outnumbered manager, bring the history: how your team's ratings compare with their measured results across cycles, and how often the forced cuts landed on your people.
- Ask HR or your skip-level manager to review voting patterns across cycles.
- Push for 360 and cross-functional evidence that the panel cannot control.
- Find one panel member willing to insist on evidence for each contested rating, so the bloc's votes have to be defended.
Employees on the losing side can ask their manager directly how calibration went and what evidence was used.
How a leader shuts it down
Appoint an outside calibration chair with independent rating authority, and let them reset outcomes visibly. Vary panel composition from cycle to cycle, and bring in 360 and cross-functional evidence. Review vote patterns by manager over several years. Appeals to vote on merit do nothing without these changes, because the incentives still reward the trade.