Setting KPIs and competencies for a fair performance appraisal

Both belong in the appraisal, and they answer different questions.

A branch manager who hits every sales target while losing half the team is a different case from one who hits the same targets and develops two successors. An appraisal that only looks at KPIs can't tell them apart.

Start from the company's goals, then go down a level at a time. Company targets become division targets, then team targets, then individual ones. Each person should be able to see how their KPIs add up to their team's.

Make each one specific and measurable. The SMART checklist, first published by George Doran in 1981, is still a useful test. In his version a good objective is Specific, Measurable, Assignable (it's clear who is responsible), Realistic and Time-related. Doran himself noted that not every objective can be put into numbers, especially for middle managers, which is one reason competencies belong in the appraisal too.

Set specific, challenging targets. Edwin Locke and Gary Latham's review of 35 years of goal-setting research found that specific, difficult goals consistently led to higher performance than telling people to do their best. The harder the goal, the higher the performance, until people reached the limits of their ability or stopped being committed to the goal.

Explain the reason for each target. In the same research, a target assigned with its purpose explained worked as well as one people set together. A target handed down with no explanation led to lower performance than one people set together.

Give people feedback during the period. Goals work better with feedback on progress than without it, so check each KPI during the period as well as at the end.

For new, complex work, set learning goals. When people are still learning a complex task, a goal to find and try out a number of ways of doing it can work better than a results target.

1. Pick a small set that applies to everyone. For example, four or five core competencies that describe how the company wants people to work. If the company already has a set of values, they can be the starting point.
2. Add a few for each job family or level. Leadership competencies for supervisors and managers, technical ones for specialist roles.
3. Describe each level as behavior. "Solves problems" is a label. "Finds the cause of a recurring production fault and puts a fix in place without being asked" is something a manager can observe and rate. Write three to five levels, from developing to role model.
4. Test the descriptions. Ask two managers to rate the same employee with them. If their ratings differ a lot, the descriptions are too vague.
5. Keep it flexible. The UK's Chartered Institute of Personnel and Development notes that well-designed frameworks can make performance expectations clearer, but that designers need to balance detail with flexibility, to avoid an approach that's overly prescriptive.

There's no single right split, but the weights should follow the role. The more a role's results can be counted and depend on the person, the more weight KPIs can carry. The more a role is about leading, coordinating or building capability, the more competencies matter.

An example, with weights chosen only to show the idea:

Whatever you choose, write it down, apply it to everyone in the same role, and keep it for the whole period. In DEUS Enhance you customize KPIs, competencies, weights, formulas, periods and measures by role and level, so different parts of the company can use different splits in the same system.

A good framework is only half of it. Fairness also depends on how the ratings are given.

• Agree first. KPI targets and the competency descriptions are agreed before the period begins, and both employee and manager can see them.
• One process for everyone. The same self-review and appraisal steps, on the same schedule, across teams and units.
• Evidence for every rating. A rating should point to KPI results or examples of behavior.
• Calibrate. Review evaluations across managers, departments or plants before results reach rewards. Two managers can rate the same performance differently, and calibration is where that gets caught.
• Look at the whole picture. Enhance's talent map plots KPI results against competency results on a nine-box grid, which shows high performers, performance gaps and development potential.

For what happens next, turning calibrated ratings into bonuses and raises, see the guide on linking performance appraisals to pay.

Plenty of companies start there, and three things make it manageable:

• Start with one level or one unit. A pilot in one division shows what works before the whole company changes.
• Begin with a workshop. Map the KPIs for each role together with the people who'll be measured and the managers who'll do the rating. When DEUS helps design a framework, it often begins with a KPI-mapping workshop.
• Review after the first cycle. Some targets will be too easy and some descriptions too vague. Fix them before the next period starts.