|
I'm gonna participate in the time-honored HN tradition of being a Debbie Downer. Dashboards suck. Not because they are inherently bad, but because of the return on investment. There are two types: 1) The dashboards that are created using blood sweat and tears...paying attention to every possible detail, trying to maximize information conveyed per square millimeter. These dashboards are works of art and are incredibly interesting when done right, but involve thousands of hours of trained developer time to get to this point. They exist for the explicit purpose of facilitating decision making procedures. And yet, in all my experience building them, the time spent maximizing information conveyed per square millimeter is a complete and utter waste, because the decisions that are facilitated by the information can, for the most part, be trivially automated and optimized using the same data sets and the same analysis that was used to create the graphic. 2) Dashboards that, due to some software package that eases their design, are incredibly easy to create...and completely uninformative. The developers of these packages think their software is great because it makes the creation of pretty graphs easy, but they leave the job of making the graphs informative up to the user. But making a graph informative is the hardest part! The myriad tweaks that are necessary to increase information density either do not exist, or they are incomplete hacks that are only possible due to a bug that might be fixed soon. I believe that the only reason they exist is to satisfy the egos of executives and micromanagers. They believe that if they have all the data at their fingertips, they can command and execute with perfection. These are the people who think you are there for one reason only: "Give me the data and I will tell you what to do". What they fail to realize is that by the time the data has been extracted, transformed, and visualized, the developer understands the data and its implications way better than anyone who only sees the high level summaries. FnordMetric has its market, and it looks like it will do very well in this niche. But it won't make your business any better. If you want software that makes your business better, look for optimization software (think Optimizely, Lokad, or my personal favorite, Gurobi)...and then empower your devs to programatically make the data-driven decisions for you. |