Research library Power laws and concentration
Performance Isn't a Bell Curve.
Across 633,263 researchers, entertainers, politicians and athletes, O'Boyle and Aguinis found the same shape: a few performers produce most of the output.
Ernest O'Boyle Jr. and Herman Aguinis, 2012
Which shape fit each of the 198 samples better
What they were looking at
Most performance systems assume a bell curve: a big middle, a few stars, a few strugglers, balanced around the average. Forced rankings, five-point rating scales and quota bands are built on it.
Ernest O’Boyle and Herman Aguinis, two management professors, asked whether real performance data looks like that. They pulled together five studies covering 198 samples and 633,263 people:
- researchers, counted by journal publications;
- entertainers, by award nominations such as Emmys;
- politicians, by how many times they were elected to the US House;
- amateur and professional athletes.
For each sample they tested whether a bell curve or a power law fit the data better.
What they found
“individual performance is not normally distributed—instead, it follows a Paretian (power law) distribution.” O’Boyle & Aguinis (2012), abstract, p. 79
It wasn’t close: 186 of the 198 samples fit a power law better. If, as their results suggest, “most performance outcomes are attributable to a small group of elite performers,” then the people who matter most to the result are the few at the top. The paper draws the line between two very different things a manager can focus on:
“the productivity of the majority of workers rather than the workers responsible for the majority of productivity” p. 108
The paper won the journal’s first Best Article Award.
What it means
If output follows a power law, the average performer is a fiction. Most people sit below the average, a few sit far above it, and rules built for the middle mismeasure both ends.
Equal territories, evenly shared leads and a bell-curve rating all assume a middle that holds most of the output. In a power law, it doesn’t.
Where it shows up in a revenue team
- Rep output is very likely heavy-tailed. A quota set on the average rep fits almost nobody.
- Round-robin routing gives the best rep the same leads as everyone else, and sends the best leads to whoever happens to be free.
- Losing one top performer can cost more than losing several average ones.
- Look at your own team the same way. Rank reps by the value of what they closed, not the count, and see how much of the total the top few carry.