Before you build a KPI dashboard read this first

Last Updated: Monday, October 1, 2012 by Johnathan Briggs

The most common mistake made by managers who are trying to build a KPI dashboard is to get metrics confused with KPI's. In every organisation there are many hundreds if not thousands of metrics and it can become very tempting to build a large dashboard that includes as many of these metrics as possible.

The problem here is we soon get information overload and it becomes hard to spot important shifts or trends in such a large amount of information. Choosing the best dashboard software alone, is not enough to ensure success.

The solution is to make sure that you have carefully worked out which of your metrics is a key performance indicator.

A key performance indicator / KPI is simply a metric that we can use as an early warning indicator to problems in the business or to track growth. For example, if we ran a support operation then a KPI might be the average resolution time of the case recorded every week. This important measure is something that we can regularly review and check the trend is going in the right direction.

So, if our case resolution time was say '4 hours' but whilst reviewing the trend data we can see that the average resolution time is slowly increasing, we might now consider taking action. So as you can see this KPI provides us with an early warning sign that allows us to take action before our customers become upset. You may well also measure many other aspects of your business for which you will want to report upon however it is important that you identify your KPI is and make sure that they are highlighted.

When building a KPI dashboard it is important that you understand who your audience is.
This very often depends on your circumstances and why you are building a dashboard.

It's possible to consider a dashboard in a very similar way to your corporate management structure, so at the very top of the tree you may have a CEO, beneath this person perhaps are departmental managers. If you then consider the typical information flow within this management structure you will most likely conclude that:

· the CEO wants to see high-level detail and only if this high-level detail raises questions will he/she want to drill down into more data.

· specific managers, like our support manager who will manage information and watch their own metrics on a daily basis, will only be expected to feed up the chain higher level KPIs.

So the CEO might only be interested in seeing the 'average resolution time' KPI and perhaps the 'number of cases' that were being submitted, but the support desk manager will be measuring many more aspects of the operation so that they can make sure that the support desk runs smoothly.

Generally it is best practice to build your dashboards in a similar layout to your corporate management structure, so typically you might have a high-level overview dashboard showing the top 8 to 10 corporate KPI's across all aspects of the business then have lower level dashboards focused on individual areas or departments within the business. These lower level dashboards should normally be owned by the managers of their respective departments and may duplicate the presentation of one or two KPI's that are shown on the overview dashboard.

If you take this approach, your dashboards will have a hierarchy, with the overview dashboard showing the most important information and the ability to effectively drill down to lower levels. So instead of forcing too much information into too small a space we structure it so if at first glance higher level KPI's raise concerns then it is quite simple for the dashboard viewer to go investigate further by drilling into the data and looking at the lower level dashboards.

Having identified your KPI's and defined a hierarchy or dashboard structure it is important that you visualise your data in the best possible way.

With so much dashboard software available many of the companies have begun to compete by producing ever more artistic charts and visualisations this has led to a crazy race between the vendor's competing on aesthetics. If you purchase dashboard software you will no doubt be presented with demonstration dashboards showing the full range of charting capabilities, and you can be easily drawn into copying some of the more beautiful charts.

There are certain chart styles that have been around for generations - they are a little boring but the reason they are still used is because they are so effective. The humble 2-D bar chart and line charts are by some distance the most elective ways to communicate 95% of KPI data.

When you start to make your chart's 3D with some unusual chart style you make your data hard to read. Dashboards should instantly communicate good and bad information to the reader and the reader should not have to work hard to interpret what they are seeing. I would strongly recommend the books and blogs of an author named Stephen Few . He has spent his career studying visual communication of data and makes many common sense suggestions about how data should be visualised the maximum effect within a dashboard or reports.

When KPI data changes your charts have a 'message' to communicate.

Most KPI's have a message behind them, perhaps we just monitor the KPI to see if it goes up to a particular threshold then we may well take some form of 'action' even if the action is simply picking up the phone and calling a colleague.

Different KPI's have different messages that you may wish to communicate so when designing any chart or data visualisation for a KPI try to think what data changes in this KPI will make the reader want to take action, then make sure your chart highlights this change. Sometimes a KPI will have multiple uses and therefore may need to communicate more than one message. In this case don't be afraid to create multiple different charts or all the visuals from the same set of data.

For example you might show the last two years monthly sales figures transposed on top of each other so that we can see how every month sales compared to the same month last year. We might also have a second visual showing a more detailed breakdown of our sales by product grouping with totals for this year to date. Whilst we might be using the same data twice, the way we present that data depends upon the message that we wish to communicate.

Before rushing to build a KPI dashboard:

1. Make sure that you have clearly defined what are your KPI's and what are just other metrics that you wish to measure.

2. Define your dashboards into a hierarchical structure that creates an effective drill down for the reader that does not overwhelm them with data.

3. When presenting any KPI or data in a visual form always consider what is the message that you are trying to make more obvious using a visual approach and make sure you choose clear chart styles that make the underlying message unmissable and don't simply pick a graphically great looking chart just because it's in your dashboard software.

Take a look at our Dashboard Best Practice Guide for advice on creating the Best KPI Dashboard for your KPIs.