kevinWhile it seems to be broadly accepted that Internet of Things (IoT) systems will connect, and extract data from, very large populations of things, a more challenging question concerns what to do with that data. The easiest, and potentially least valuable answer is to employ visualization tools that allow humans to see the data in a more comprehendible manner. This is often as far as many businesses have gotten with IoT, which has understandably raised frustration levels among corporate executives. Simply visualizing complex data does not, by itself, yield better business outcomes.

While IoT dashboards might be nice to have they suffer from two fundamental problems. The first is scale. The implicit assumption with dashboards is that humans will inspect the dashboard, see events requiring attention, and act upon them. But whenever we have humans in the equation it automatically means the system won’t scale. Keep in mind that for many businesses the expectation
is that tens of thousands, possibly hundreds of thousands of devices while be generating operational and environmental data at a fairly high frequency. Even skilled humans cannot keep up with that volume of data, no matter how clever the visualization techniques.

The second problem is pattern recognition. While humans are unquestionably highly skilled at pattern recognition, with complex IoT systems the patterns in question may encompass considerable time. For example, data analytics may have determined that if events a, b, and c occur within 5 minutes of each other and 30 minutes after event d, a failure is likely to occur. Humans have difficulty perceiving these kinds of patterns and, again, sophisticated dashboards are unlikely to help.

The most fundamental aspects of properly designed IoT systems, apart from basic device connectivity, are (a) rules that can be applied to complex data sets in order to seek out relevant patterns, and (b) actions that can automatically be taken in order to prevent or minimize adverse events. These functions are not dependent on human interaction and can therefore take place continuously, at high speeds, and at scale.

So, sure, go ahead and build yourself a dashboard. But don’t expect to use it for anything other than showing off to management.