3 minute read – Posted by – June 21, 2018

How to filter noisy performance data

Performance management—it seems like all companies do it but for what purpose?

As it stands currently, the general purpose of performance management is to obtain feedback in order to differentiate chance vs skill in employee performance, recognize employee growth and set new measurable objectives.

It is, however, difficult to make comparisons between employees that aren’t within identical situations.

Let us begin this discussion by assuming all employees have equal abilities to perform to exceptional standards. Despite this assumption, our measure of employee performance of choice will likely be noisy.

Consider the following example in a normal distribution:

At any given time and any level of effort, results for measurable outcomes of employee performance can vary and lie outside their control. For instance, the efforts of team members or a thriving organization. The key here is to differentiate skill from luck.

In some instances, it’s easy to distinguish between skill and luck:

In this particular distribution, a high efforts distribution is much closer to good outcomes compared to lower efforts. This makes it easy to infer that high levels of effort from employees will likely result in good outcomes and vice versa.

While this simple distribution is nowhere from perfect, we can see that under these circumstances employee performance is easy to measure and straightforward feedback can be provided.

Screen Shot 2016-04-01 at 9.52.49 AM

Credits: Cade Massey, Wharton School of Business

In the real world, it is far more difficult to distinguish between skill and luck:

In this case, while higher efforts do correlate to a higher likelihood of good outcomes, the chances of an employee performing exceptionally by chance and skill is mostly indistinguishable. And vice versa, when poor outcomes results, this could very well be a result of chance and not the employee’s lack of ability to perform.

All of this stems from the noise in the data that we’re using to measure employee performance.

Screen Shot 2016-04-01 at 10.05.37 AM

Credits: Cade Massey, Wharton School of Business

All of this said bottom line is, the inferences that can be made between chance and skill in employee performance management and objective outcomes is never straightforward.

Whether you’re more concerned about how your performance is being managed or hoping to learn more about managing performance, we’ll be looking at how to find a signal in the noisy data.

Stay tuned for more next week!


Also published on Medium.

May Chau

May is a Content Strategist contributing to the improvement of modern performance management at 7Geese. Connect with her via may@7geese.com