This column was written by Florian Kahlert, Managing Director, North America, of GfK’s Digital Market Intelligence (DMI) group.
A dashboard, as the name suggests, should give the user a very quick readout on the status of something. Think of a car dashboard: You have a bunch of lights, gas and temperature gauges, and maybe a tachometer. The temperature gauge usually stays around the middle once the car is warmed up; if it moves into the red zone, you know you should stop the car and investigate why.
Similarly, business dashboards can tell you clearly what is (or isn’t) happening. They may even pull data points together to show you a trajectory or trend. But to understand the underlying causes requires further analysis. And this is where I have issues with dashboards – particularly the marketing variety.
“Does GfK have a dashboard?” is an increasingly common question in research pitch meetings these days. Dashboards can become addictive in the same way as Facebook pages; users can get caught up checking for fresh data and forget about the bigger picture. Looking at a campaign on a minute-by-minute level tells you little about how your advertising will fare in the long run – and it certainly can’t give you recommendations on how to proceed.
Dashboards can also give their users a false sense of control over the process at hand, simply because one can look at it all the time. No matter how sophisticated your telescope may be, it is still just observing.
Finally, people can get hung up on the flashy coolness of how dashboards look and forget about what the data actually means. Form should never trump substance; it should simply support making the data easily understandable.
Now, I am sure that your company isn’t making these mistakes. But, believe it or not, many are. It is easy to forget that, while you can address symptoms based on dashboards, it takes initiative and energy to locate and fix the root causes – just as it is necessary to open the hood of a car to figure out if the radiator is busted or the waterpump gave out.
With this in mind, here are some essential precepts for developing dashboards with actionable information in mind.
- Be sure that your dashboard provides context along with the data. How deep are you into the campaign’s run? How close are you to your minimum sample size? Knowing that you have 50 datapoints is one thing; but knowing if that constitutes 100% or 1% of what you are looking for is totally different. Simply providing such context – through customizable dashboard elements — can tell the user how seriously to take whatever he or she may be seeing in readouts.
- Allow users to drill down below the surface. Going beyond the “dashboard” to provide a user interface that allows data analysis (in our case, filters and cross-tabulation options) that allow you to understand the causes for the most common use cases. For the rest, provide the ability to download respondent-level data sets so clients can perform their own custom analysis
- When all else fails, be sure consultation is available. Even with dashboards and drill-downs, many clients are not just asking for data (the what), but also which actions they might want to take (the what now?!).
Dashboards are great tools for identifying variations and trends – but they are never meant to be cure-alls. A dashboard is just a starting point, a sign showing the way toward areas for further analysis. Understanding why trends and situations are happening means going to the next level – and a good dashboard should make that process easy.