How to set realistic targets for your KPIs?

One of the most difficult parts of Performance Management is to come up with realistic targets for your KPIs. I have seen many people confuse company goals with KPI targets.

An example will illustrate this: Lets say you launch a new product and the retention of this product. The company goal might be to have 120.000 items sold 24 months after the launch.

How do you measure if you are on track to reach this goal?
Well the people that confuse goals and targets define an indicator called: “Total number of items sold” and set the target for this indicator at 120.000 and start measuring every month. If you do this, your dashboard will most likely be dark red for the next 20-28 months. The interpretation of anybody looking at the dashboard will be that business is going bad. And this might not be the case.

Conclusion: we need to transform the goal into a target in relation to its measurement frequency.

If we do this linear: 120.000 / 24 = 5000 items sold each month. Oops, most businesses are not linear. In this example items sold is heavily influenced by the marketing push, viral marketing, availability in stores etc.

In other words, creating realistic targets is not so easy. Targets must be related to the frequency of measurements and need to take seasonal influences and external influencers (cause and effect) into account. Hmm, this starts to get really complicated…

Sometimes we need to think out of the box and see if we can learn something from another industry, in this example financial analysis of stocks.

When analyzing stocks, economists use technical trend analysis to analyse stocks. The idea is that there are too many variables that influence a stockprice: The company performance, the market performance, the economy, consumer trust, interest rates, the weather, the time of year, sentiment etc.
The outcome of technical analysis is to determine “resistance” and “support” levels for stocks where resistance is the next “goal level” for the trend and support level the “risk level”.

The interesting thing to take away is that it is possible to define new goals and risks based on historical scores.

With some KPIs this method could be used to set realistic targets. Stacey Barr pointed out in a previous blogpost that a KPI needs to have about 20 measures to become statistically relevant.
This is why you should prefer KPIs that you can measure every day or at least every week. You should also prefer running periods (last 30 days) over closed periods. (last month) as running periods are more statistically viable.

Once you have a good trend of 20 measures of running periods to filter out seasonable and weekend influences, you could apply the technical analysis principle in order to find resistance and support levels for your KPI. Most likely you will find that this is easier and more reliable then trying to model all the cause and effect relations in your business.

And, more importantly, by looking at historic performance you are create realistic targets for your employees to reach. After all, it is quite motivating when a dashboard turns green once in a while!