Thursday, July 10, 2008


Twenty-seventy-ten. I hope those numbers hold no special meaning to you. I occasionally see those numbers on white boards around the office and overhear them in conversations.

20-70-10 is a technique that some high level managers advocate for driving a high performance organization. It's part of the Jack Welch GE management system. In short, the system is set in place to differentiate and rank all employees and systematically remove the bottom ten percent of the group and generously reward the top twenty percent. The idea is that the best people in the organization will stick around and you drop the dead weight.

I've made a little chart depicting the theory. In the chart I have an organization that consists of people that I ranked by strength, 1-10 and also relative strength within the organization. For example, after the first iteration those employees with the bottom ten percentile of strength is 1 were removed from the pool and a set of new employees with a strength of the mean strength of the group, 5.5 is added. This is purely an abstract model of the system, however certain trends appear quickly. Within five years the weakest group are at the same strength that was at the middle of the pack at the beginning of the iterations.

There are also some other interesting projections in this model. Look at the top. It's virtually unaffected. The top 30% can stay there through the whole exercise without making any adjustments. The 60-69% can stay in that position without adjusting their strength for 5 years.
The strength of this system is that it does purge the weakest people from the system. There's no doubt about that.

What my model fails to capture is the forces at the top. The people in the top 20% are certainly motivated to compete to stay up there. Providing that kind of motivation though isn't cheap. If the carrot were to motivate the top 20% it will need to be substantial.

To offset some of that expense, the middle 70 and, even more, the bottom 10 will enjoy much less incentive than they may feel they deserve. Attrition becomes an issue, certainly those who are calibrated at the bottom are forced out, but there are others who will leave because they feel that they are not being fairly recognized for their contributions. Are these people so bad that by terminating them the organization is better off? How does a system like this deal with valuable people who are lazy?

Take the case of the person who comes in at 21. One person shy of the beautiful people. What is her reaction to this assessment? Did she contribute less than the top 20? How can she be certain that the calibration system is accurate or equitable? She really can't.

What are her options? Well, she can try harder and try to break into the top 20%, but isn't there another option? This isn't a closed system. There are other options outside of the organization. That has to be a consideration and a legitimate risk to lose excellent people.

A system like the 20-70-10 is good for cleaning house. It will clear out the weakest employees. This isn't necessarily a bad thing to purge the weakest people out of an organization when those people are considerably weaker than those who are brought in to replace them.

The hiring assumption of this model is that on an aggregate level only those who are as strong as the mean of the group are hired. This system raises the hiring standard on a logarithmic scale. Each subsequent year of the system raises the bar less and less.

I do not believe that a 20-70-10 system should be sustained for an extended period of time. Given ideal circumstances it starts to lose its value after a few iterations.

When you factor real world situations into the model it does not improve its sustainable effectiveness. Consider a situation where people within the organization are laid off. This happened in my own organization. About 25% or my colleagues were laid off for cost cutting. This happened independently of the 20-70-10 purging process. Those who were laid off were within the bottom 50% of the organization so when it came time to let the bottom 10 know who they are, we had people who were possibly in the lower third of the company at the start of the year ending up in the lowest 10 at the end.

Occurrences like the one outlined above undermines peoples senses of security. When people worry about termination and their ability to meet their needs, they aren't productive.

What organizations in the field are finding is a 20-70-10 system can add a great deal of value for a few iterations. It is excellent at creating a culture that does not tolerate underperformers, but if allowed to run unchecked the system loses its value after a few iterations.

No comments: