What it is and why it spread
Forced ranking requires a manager to place a fixed proportion of their team into each performance band: a set percentage as top performers, a set percentage in the middle, and a set percentage at the bottom, often with consequences attached to the bottom band.
It was adopted for a real reason. Left to themselves, managers rate generously. Ratings drift upward, differentiation disappears, and a distribution where almost everyone exceeds expectations tells the organisation nothing and makes pay and promotion decisions unsupportable.
Forced distribution guarantees differentiation. That is its genuine attraction and it is not a trivial one, because the alternative many organisations actually have is a rating system that has stopped discriminating at all.
The question is whether guaranteed differentiation is worth what it costs, and the accumulated experience of organisations that ran it is that it is not.
The small numbers problem
A distribution is a statement about a large population. Applied to a team of six, it is a statement about six people, and it will frequently be false.
If ten per cent must be rated at the bottom, a team of six either has nobody there, which breaks the rule, or has one person at seventeen per cent of the team, which overshoots it. Either way the manager is being asked to produce a shape the data does not have.
Worse, a genuinely strong team of six contains no poor performer, and the rule requires one to be nominated anyway. The person nominated is being rated against a quota rather than against a standard, and both they and their colleagues generally know it.
This is the mechanism by which the system loses credibility fastest. Everyone can see that the bottom rating in a strong team means something different from the bottom rating in a weak one, and once that is visible the ratings stop carrying information.
What it does to behaviour
| Intended effect | Actual effect |
|---|---|
| Differentiates performance | Differentiates within teams, not across them, so ratings are not comparable |
| Identifies poor performers | Identifies the weakest person present, who may be performing well |
| Motivates improvement | Motivates competition with colleagues, since the bands are zero-sum |
| Rewards top performers | Encourages strong performers to avoid strong teams |
| Raises standards | Discourages managers from hiring strong people who would crowd their curve |
The collaboration effect is the most damaging and the least visible. Where bands are fixed, helping a colleague succeed can move you down. Nobody says this out loud and everybody understands it, and the behaviour adjusts quietly.
The bottom two rows describe a slower failure. Managers learn to protect their distribution by avoiding strong hires, and strong performers learn to avoid teams full of other strong performers, which is precisely the opposite of the intended sorting.
Why organisations abandoned it
A large number of well-known adopters have moved away from forced distribution, and the reasons given are consistent.
- The ratings stopped being comparable across teams, so the data could not support the pay and promotion decisions it existed to inform.
- Voluntary attrition rose among people rated in the middle, who concluded the system had already decided about them.
- Managers spent the cycle negotiating quotas rather than assessing performance.
- Legal exposure increased where the pattern of who landed in the bottom band was uneven across groups.
- It became a recruiting liability once candidates knew the organisation ran it.
The fourth is worth stating carefully. Forced distribution does not create bias, and it does concentrate its consequences: a system that requires a bottom band each cycle produces a documented record of who was placed there, and any uneven pattern in that record is both visible and hard to explain.
What to do instead
The problem forced ranking addresses is real, so abandoning it without replacing it returns the organisation to rating inflation.
Calibration is the usual answer. Managers assess against a standard, then meet to compare their assessments and align on what each rating means, with evidence. The distribution that results is whatever the evidence supports rather than a shape decided in advance.
That preserves the differentiation forced ranking was protecting, without requiring anyone to nominate a colleague to fill a quota. It also produces something forced ranking cannot: an explanation of why a rating was given that the employee can be shown.
- Define what each rating level actually means, in observable terms, before the cycle rather than during it.
- Calibrate across teams so a rating means the same thing in both, which is the real comparability problem.
- Report the resulting distribution and look at it, without mandating it. A distribution where nearly everyone exceeds expectations is a signal to act on, not to enforce against.
- Separate the performance conversation from the pay decision, since combining them makes both worse.
Where a distribution is reported and stubbornly implausible, the fix is standards and calibration discipline rather than a quota, because a quota does not make the assessments more accurate. It only makes them differently shaped.
Frequently asked questions
What is forced ranking?
A system requiring managers to allocate performance ratings to a predetermined distribution, so a set proportion falls into each band regardless of actual performance. It is also called forced distribution or stack ranking.
Why did organisations adopt it?
To counter rating inflation. Left alone, managers rate generously, differentiation disappears, and a distribution where almost everyone exceeds expectations cannot support pay or promotion decisions. Forced distribution guarantees differentiation, which is a real attraction.
What is wrong with forced ranking?
A distribution is a claim about a large population, and applied to a team of six it will frequently be false. A strong team is required to nominate a bottom performer anyway, everyone knows the rating means something different there, and the ratings stop carrying information.
How does forced ranking affect collaboration?
It makes helping a colleague individually costly, because the bands are zero-sum and someone else moving up can move you down. Nobody states this and the behaviour adjusts anyway.
What should be used instead?
Calibration. Managers assess against a defined standard, then meet to align on what each rating means with evidence. The distribution is whatever the evidence supports, and unlike a quota it produces a reason for the rating that can be shown to the employee.
How Engage supports calibration instead
Engage brings proposed ratings and the evidence recorded against them into one view, so a calibration session compares what people did rather than negotiating quotas. The resulting distribution is reportable by team, manager and group, which lets an organisation see inflation or an uneven pattern without mandating a curve to correct it.
See calibration in Engage