Here is a belief that runs half the appraisal cycles in the country: in any large group, a few people are brilliant, a few are poor, and most sit in the middle, so ratings should be spread the same way. Hence the familiar rule that roughly 10% get the top rating, 70% the middle and 20% the bottom, or some local variation of it. The shape on the slide is a bell. The claim underneath is that this is how performance naturally looks.
It is worth taking that claim apart, because the answer decides whether the rule is a sensible tool or a source of damage.
Where the bell came from
Many things in nature do follow a normal distribution: height, for example, or measurement errors. Because they do, it feels natural to assume that human output does too. Forced distribution, also called forced ranking or the vitality curve in the version General Electric made famous under Jack Welch, takes this idea and turns it into a quota. Managers must place a set share of their people in each rating bucket.
The practical attraction is easy to see. Managers who are all “generous” give everyone 4 out of 5, and then the increment budget cannot differentiate anyone. A quota forces conversations and protects the budget. That is a real problem, and the quota does solve it. I would not pretend otherwise.
What the research says about the shape
In 2012, Ernest O’Boyle and Herman Aguinis published a paper in the journal Personnel Psychology titled “The Best and the Rest: Revisiting the Norm of Normality of Individual Performance”. They examined a large number of datasets covering researchers, entertainers, politicians and athletes, using objective output counts instead of manager ratings, and found that performance in these groups was often better described by a power-law shape than a bell. A small number of top performers produced a disproportionate share of the results. The finding does not apply uniformly to every job, but it does challenge the assumption that the bell is the default.
Think of it in a sales team. If 5 of 100 salespeople bring in a third of the revenue, forcing only 10% into “top” and capping that bucket is not neutral. It may lump people whose contribution differs hugely into one box.
Small teams break the maths
Even if the bell were true across the whole company, it need not be true of your team. A statistical distribution describes a large population. Take a team of ten with a rule of 20% in the bottom bucket. Two people must be rated “below expectations”. What if all ten are hired well, trained well and doing solid work? Two still take the label. What if a team of eight has a rule of 10% at the top? That is 0.8 of a person, so managers round, argue and negotiate. The rating then reflects the negotiating strength of the manager, not the contribution of the employee.
That is why most companies that use quotas apply them to a calibration pool of a few hundred people, not a single team. Calibration helps, but the quota still shapes the discussion: managers arrive already thinking about whose name goes into the bottom slot.
The side effects people complain about
- Collaboration suffers when a colleague’s gain is your loss. Why would you help a peer if only one of you can have the top rating?
- Good teams get punished. A strong team must still hand over its share of bottom ratings, and good people leave.
- Managers game the system, with rotation of the “bad rating” across years, or by moving weak performers to other teams before calibration.
- The bottom bucket becomes a hiring-and-firing mechanism, which is not what an appraisal is for.
Several large employers have moved away from it. Microsoft ended its stack ranking in 2013, and Accenture announced in 2015 that it would drop annual ratings and rankings altogether. Others, including many Indian IT services firms, keep some form of normalisation. The direction of travel in the industry has been away from rigid quotas, though a lot of companies keep softer versions.
If not a bell, then what?
The honest alternative is not “rate everyone high”. It is clear role-level expectations, calibration sessions where managers defend ratings with evidence, and guidance ranges instead of fixed quotas. A guideline such as “we normally see 10 to 20 percent in the top bucket across the company; explain if your team is far outside” gets the benefit of discussion without compelling a loser.
A quota has a place when a company has a real rating-inflation problem and no better habit. Treat it as a temporary discipline, though, not a theory of human nature. People are not shaped like a bell. Budgets are what force the sorting, and it is better to say so openly.
Figures and names in the examples are illustrative.
Leave a Reply