Share:

We all want KPIs to drive the right behaviour in employees. But there is a good reason why it’s so hard to do.

When we say we want KPIs to drive the right behaviour, what we really mean is that we want people to use KPIs in a way that improves the organisation’s performance, holistically. We don’t want:

• A single KPI to hit its target at the expense of other KPIs, like hitting a cycle time target at the expense of safety or workmanship
• A KPI to improve numerically while the result it is supposed to measure gets worse, like reducing customer complaints by redefining a complaint while more customers leave to competitors

The behaviours we don’t want KPIs to drive are called ‘gaming’. It’s when the calculation, data or visualisation of a KPI is modified to make it look positive, or when business operations are performed in a way to generate positive data for a KPI.

But we’ve got it back-to-front. It’s not the KPIs that are driving gaming behaviour.

Even good KPIs can be gamed.

It’s true that there are many trivial measures with aggressive targets that can only be achieved by gaming. Any kind of quota easily drives the wrong behaviour. The
clichéd example is sales people hitting quotas for sales calls by calling anyone, not ideal customers. And the consequence of gaming these quotas is that sales conversion drops.

Lost Time Injury Frequency Rate (LTIFR) is a common measure in organisations where people have exposure to physical harm. Think energy generation, transport or construction. LTIFR quantifies how often injuries happen that result in lost time off work. It’s a useful measure and an important measure. But it can be gamed.

If there is pressure to hit a target for lower LTIFR, a common way to game this is for people to stop taking time off if they have an injury at work. For the safety and well-being of employees, and the purpose of this measure, that’s the wrong behaviour.

Clearly, it’s not the goodness of the KPI that drives behaviour.

It’s the measurement process that drives behaviour, not the KPIs.

A better chisel does not produce a better carpenter. Better KPIs won’t drive better behaviour. Only the process of performance measurement, and the performance culture it supports, can do that.

There are several reasons why a measurement process can lead to good measures like LTIFR being gamed, such as:

• Targets are set without human sensitivity
• Targets are expectations, not intentions
• Accountability is based on hitting targets

If people believe they will be judged or blamed for missing a target they don’t have the capability, resources or authority to reach, they will do whatever they can to make it look like the target is reached.

Focusing on the design of better KPIs will not fix this underlying cause that drives the wrong behaviour.

Steps your measurement process must have to drive the right behaviour.

We do need good KPIs so that people know the results that matter and need to be improved. And a good measurement process will generate good KPIs. But we’ll only get the right behaviour when the entire measurement process supports the right behaviour.

In PuMP we do measurement a very specific way to support people to use KPIs to improve organisational performance, holistically. Some of the important techniques PuMP uses include:

• Accountability for KPIs is defined as monitoring, interpreting and responding to the KPI, not hitting targets
• KPIs are developed by the people who will use them to improve performance
• No KPI is selected without deliberate consideration of its unintended consequences
• Every KPI is linked to related KPIs, to ensure they can all be used to get an holistic understanding of organisational performance
• We use KPIs to help us continually improve our processes to reach for targets, not to hit them.

When these techniques are consistently practiced, they develop cultural norms (we also call them memes of performance measurement) that drive the right behaviour as a natural and easy consequence. Some of these are:

• Measure to learn and improve, not compare and judge.
• A performance culture grows with buy-in, not sign-off.
• Assumption is the mother of all poor performance measures.
• Interpret patterns, not points; signals, not noise.
• Fundamental improvement happens through leverage, not force.

Performance measurement only works when it dignifies people. But overwhelmingly, traditional approaches to performance measurement do not dignify people; they judge and blame and punish people. And often for results outside their control. Get your measurement process working right, and the right behaviour will be a natural consequence.

A better chisel does not produce a better carpenter. Better KPIs won’t drive better behaviour. Only the process of performance measurement can do that. [tweet this]