Most performance management systems set goals at the beginning of the year and determine variable compensation by rating accomplishment of those objectives.
These evaluations typically are considered in lengthy “consensus meetings” in which managers discuss the performance of hundreds of people in relation to their peers – sometimes called “stack ranking,” or more cynically “rank-and-yank.”
These year-end ratings don’t provide “in-the-moment” and “real-time” feedback about actual performance as it happens, so may be less useful in improving performance.
Assessing skills produces inconsistent data based on raters’ own skills in that competency and the value they attach to each performance objective, leading to unconscious bias.
This risk to performance rating validity was demonstrated by Drake University’s Steven Scullen, Michael Mount of University of Iowa, and Korn Ferry’s Maynard Goff, who considered 360 degree performance evaluations by two bosses, two peers, and two subordinates for nearly 4500 managers.
They found that three times as much rating variance was explained by individual raters’ idiosyncratic evaluation choices, rather than actual performance.
Sources of bias include halo error, leniency error, and organizational perspective based on current role, suggested by SUNY’s Manuel London and James Smither of LaSalle University, and validated by Scullen’s team.
These findings led the researchers to conclude “Most of what is being measured by the ratings is the unique rating tendencies of the rater. Thus ratings reveal more about the rater than they do about the ratee,” replicating similar findings by University of Georgia’s Charles Lance, Julie LaPointe and Amy Stewart.
To mitigate these biases in Deloitte’s performance management system, Ashley Goodall of Deloitte Services LP engaged Marcus Buckingham, formerly of The Gallup Organization, to analyze existing practices and develop an empirically-validated approach.
Goodall and Buckingham calculated the total annual hours required to conduct performance ratings using the existing process and found that managers invested 2 million hours a year.
This finding confirmed that one goal in revising the process was to increase speed and efficiency.
In addition, Goodall and Buckingham sought to increase the meaningfulness of performance management by focusing on discussions about future performance and careers rather than on the appraisal process.
They concluded a performance management system should be characterized by:
- Reliable performance data, controlling for idiosyncratic rater effects,
- Speed to administer,
- Ability to recognize performance,
- Personalization: “One-size-fits-one”,
- Considering actions to take in response to data,
- Continuous learning and improvement.
Deloitte conducted a separate controlled study of 60 high-performing teams including almost 1300 employees representing all parts of the organization compared with an equal number of employees from an equivalent sample to determine questionnaire items that differentiate high- and lower-performing teams.
They found that performance and related compensation allocations could be more accurately based on managers’ statements about their intended future actions toward each employee rather than asking about team members’ skills.
Several items accounted for the vast majority of response variation between top performing groups and others, particularly “At work, I have the opportunity to do what I do best every day.”
Business units whose employees said they “strongly agree” with this item were substantially more likely to be more productive, earn high customer satisfaction scores, and experience low employee turnover.
Other powerful predictors of performance were:
- I have the chance to use my strengths every day,
- My coworkers are committed to doing quality work,
- The mission of our company inspires me.
Deloitte’s revised performance management system asks team leaders to rate four items on a 5-point scale from “strongly agree” to “strongly disagree” or yes-no at the end of every project or once a quarter:
- Given what I know of this person’s performance, and if it were my money, I would award this person the highest possible compensation increase and bonus [measures overall performance and unique value],
- Given what I know of this person’s performance, I would always want him or her on my team [measures ability to work well with others],
- This person is at risk for low performance [identifies problems that might harm the customer or the team],
- This person is ready for promotion today [measures potential].
These responses provide a performance snapshot that informs but doesn’t completely determine compensation.
Other factors include project assignment difficulty and contributions other than formal projects, evaluated by a leader who knows each individual personally or by a group considering data across several groups.
In addition, every team leader prioritizes once-weekly “check-ins” with each employee to ensure that priorities are clear and progress toward them is consistent.
Goodall and Buckingham opined that “radically frequent check-ins are a team leader’s killer app to recognize, see, and fuel performance,” in addition to using a self-assessment tool that identifies each team members’ strengths and enables sharing with teammates, team leader, and the organization.
These three “interlocking rituals” of the weekly check-in, quarterly or project-end performance snapshot, and annual compensation decision enable a shift from retrospective view of performance to more “real-time” coaching to support performance planning and enhancement.
Deloitte’s approach seeks a “big data“ view of each person’s organizational performance and contribution rather than the “simplicity” of a small data view summarized in a single stack-rank number.
-*How do you develop a “Big Data” view of people’s performance?
-*How do you enable continuous, “in-the-moment” performance feedback instead of once-a-year retrospective view?
- Comparative Rankings May Reduce Gender Bias in Career Advancement
- “Social Accounts” as Pay Substitutes = Lower Pay for Women
- Do Unintended Consequences of Forced-Ranking of Employee Performance Outweigh their Short-Term Benefits?
- Biases in Unconscious Automatic Mental Processing, and “Work-Arounds”
- Overcoming Decision Bias: Allure of “Availability Heuristic”, “Primacy Effect”
- Detect and Mitigate Decision Biases
- Male Peer Raters Discount Women’s Expertise in Science, Engineering