Performance and Rewards: Differentiate Only When It’s Worth It

Should you encourage managers to differentiate performance and rewards among workers? The typical assumption is that more differentiation is better. Merit-based rewards are seen as a hallmark of a “high performance culture.”

• 91% of companies linked pay to performance in 2011, up from 78% in 2009.

• 89% of companies differentiated performance using numerical ratings in 2014­-2015.

Yet, some leading organizations are adopting performance management systems that reduce or remove ranking, numerical ratings and performance categories. GE recently announced changes to its famous annual forced-ranking performance management system. Microsoft abandoned its forced­ranking system in 2013. Expedia, REI, Adobe, Deloitte and Juniper Networks also gained headlines for moving to rating­-less performance management systems.

Should we rethink the idea that differentiating performance and rewards is always good?

Research Shows That Differentiation Doesn’t Always Pay Off

Economists have reviewed research on whether wider individual differences in total pay relate to better performance, and the evidence is inconclusive. Some studies find a positive relationship, others a negative relationship (especially in collaborative situations), some a hump shape with a “sweet spot” in the middle, and others studies a u­-shape with a “sour spot” in the middle.

One study of major league baseball teams between 1985 to 1993 found that both individual player and team performance were lower when there was greater pay differentiation. Other researchers have found similar results in baseball and hockey. The impact of differentiation on retention is however more uniformly negative.

Assuming that differentiation is always beneficial, or at least does no harm, is not supported by research. How can you differentiate where it does the most good and the least harm?

Performance Differentiation As Total Quality Management: Three Vital Conditions

Performance and pay differentiation can be viewed through the lens of total quality management, where a fundamental premise is to remove variation unless it adds value. Reducing variation in performance and pay can be consistent with this premise. Differentiation in any process, including worker performance and pay, should only be attempted if three conditions are met:

1. Differentiation makes a pivotal impact on valuable outcomes

2. Differentiation can be achieved with sufficient precision, reliability and accuracy

3. The cost of measurement does not exceed its value

Regarding condition #1: The “return on improved performance” or ROIP from good to great is not always significant, as with commercial airline pilots, where the key is to meet a high performance standard, but not to show who is “top gun. ” Most organizations need financial reporting accountants who get their reports done accurately and on time, but do not need them to compete to show who is most innovative.

Don’t assume that performance increases in every job are valuable. Sometimes “good enough” is the optimum goal.

Regarding condition #2: Performance ratings are prone to bias and unreliability. One research review found only between 8% and 32% of the variation in performance ratings is due to individuals and their achievements. People’s observations and judgments can be very inaccurate and performance appraisal research shows people display biases and take shortcuts that interfere with their judgments. Supervisors also have many agendas and aren’t always motivated to rate performance accurately. Factors beyond the control of the employee can also influence their performance, leading to errors in performance judgments. Finally, efforts to improve performance ratings through scale improvements, rater training, calibration processes, 360 degree feedback, and forced distributions have not met with much success.

Don’t assume performance can be accurately observed.

Regarding condition #3: The costs of differentiation are significant. At a recent conference several performance management professionals in large companies said that reducing performance differentiation saved millions of dollars per year in direct system costs and worker and manager time. Systems to measure and differentiate performance are costly to put in place, especially if there are technology applications involved. The time and effort required of employees and supervisors is significant. Then, there are the billions of dollars that are riding on performance measures. The cost and management attention required to manage a pay­-for­-performance system is enormous.

Don’t assume the costs are trivial.

Differentiate Performance and Rewards Only When It Definitely Pays Off

Merit­-based systems that differentiate performance and rewards are undoubtedly valuable when they work, but leaders often strive for differentiation without question. Even when organizations modify their systems to remove the requirement to rank or rate employees, managers still assume they are expected to differentiate rewards, so they use factors other than performance (such as judgments of potential, agility, or cultural fit) that may be even more problematic.

Yet, research suggests that these three essential conditions are met only in certain circumstances. When they are not, it may be better not to differentiate performance and rewards, or to differentiate only very broadly. Juniper Networks not only abolished forced rankings, it seeks simply to have what it calls Best Talent. More than 97% of its employees are considered Best Talent, and Juniper’s talent management efforts focus on putting the right capabilities in the right places to achieve its business goals.

Perhaps a better approach is to make differentiation the exception not the rule, and use it only where it clearly pays off.