KPI Traffic Lights - 3 ways to highlight the real signals in your performance measures

Traffic lights - the decoration of rigor for performance dashboards and reports. Have you gotten carried away with the decoration, which rigor with? Take a look at these four common approaches to traffic lights, and see if you have some room for improvement.

Approach 1:% difference from month to month

When this month is 10% worse than last month, the traffic light turns red. When 5% worse than last month, the traffic light turns yellow. When it's 10% more than last month, the traffic light turns green. Obviously, this approach works for periods of time of one month, and the cut-off other than 10% and 5%.

These traffic lights encourage us, generally, to ask questions like "what caused such a big difference?" In turn, these questions will encourage, usually to find a way to explain the difference. If we're smart, we've added a comment to the extent of saying that the performance difference is due to something outside of our control. If you're not so smart, we will get a different explanation each month, and have a long list like Santa Claus' improvement projects.

There is no advantage I can see this approach to the traffic light. It tends to encourage us to react instinctively to the data, tamper with the business processes or blaming something that you do not need to do anything. Time is wasted chasing problems that there are problems that are missing and there.

Approach 2: up and down, good and evil

When some extent the performance values ​​increase, it is a good thing (like revenue performance, satisfaction and on-time). There are others whose values ​​decrease and is a good thing (like the rework cycle time, and pollution). Combine this with the fact that there is a change upward or downward change in the values ​​of current performance and produces a complex range of traffic light signals to address: the change up is good, the change to up is bad, change down is good, change down is bad. This "solution", probably the result of confusion that erupted when the arrows up and down were chosen as symbols of traffic lights.

When we resolve the confusion, these multi-faceted traffic lights encourage us to ask questions like "what's behind the trend?" and the trend is concluded from maybe 3 consecutive data points. Marginally better than approach # 1, and just.

Any system of traffic lights you move away from point to point (compare the essence of the method # 1) is a step in the right direction. But we still have to risk drawing the wrong conclusions from trend analysis, which is based on data not yet enough to be valid. And it does upwards and downwards matter nearly as much good and evil?

Approach 3: Signals statistically valid

Statistical process control is a method of analysis that discerns the variation that is typical of that change means change has occurred. And 'how to filter signals from noise, something that the other two approaches are not (we assume that any arbitrary difference is a signal, regardless of the size of the typical differences in time). The signals are defined by a set of rules that test the probability that the difference is just due to normal variability (no change) compared with atypical variability (change). The signs are sudden changes in performance, gradual changes in performance and instability in terms of performance.

When our attention is shifted from point to point variations to the models to change over time, we ask questions like "what caused this change in performance that occurs at that time?" and "because the performances so chaotic and unstable?" and "what we need to focus on improvements to improve the overall average level of performance? '.

These questions seek the root causes, not causes symptoms. They lead us to find solutions that do not limit themselves to setting the performance of next month, but fundamentally improve the level of basic services further into the future .......