In these days of increasingly powerful ways to process higher and higher volumes of data the challenge still remains: how do you communicate the insights? This hasn’t really gotten any easier. Many BI/Analytics tools have built-in “chart choosers” or wizards, but generally they aren’t particularly good at helping decide how to communicate these insights.
The main reasons for this inability to make good recommendations are because these automated / assisted approaches understand far too little about the data to know which are good suggestions and which aren’t. This understanding goes way beyond traditional metadata, although at least that is a start.
Surprisingly perhaps, it isn’t rocket science to narrow down the possible chart types to those which are more, and less, appropriate. There are a couple of key factors to consider. The most important one is to understand which of six (yes, only 6) business questions the business person wants answered.
One of these occurs with much more frequency than all the others. And executives tend to care most about just two of them. Identifying which is the most important question to the business user/community usually surfaces during the first few minutes of conversation for requirements gathering. (Hint: it is a complete waste of time asking business users detailed questions about what exactly they want to see because they don’t - and can’t possibly - know what they want until they see it. The solution for showing them what they want is rapid cosmetic prototyping, in a few hours to a day or so, but that is a separate subject.)
Back to choosing chart types…the other key factor to understand is the volume profile of the data. By this I mean the number of categories/members in dimension levels, not the number of facts (rows) to be processed. Armed with only these two initial pieces of information an experienced designer can quickly limit the possibilities to those which are effective at communicating insight in a given business scenario. There are also about 4 or 5 other questions which help in refinement, but these generally tend to be secondary.
By limiting possibilities, I mean reducing the number of chart types down from the 15 (or so) major types and the ~70 variants, to a small number. The devil is in the detail, it isn’t the major chart types that are the challenge, it is understanding the purpose and appropriateness of the variants.
When it comes to providing dashboards, particularly strategic ones for executives, and tactical ones for managers and professionals, it is surprising how much can be communicated with fairly straightforward charting. I am a huge fan of (information) visualization, but that is more than just charting, it is about telling and explaining (business) stories and requires additional skills, as well as some more flexible technology. In fact, my favourite visualization was designed by Charles Minard in 1869, using only pen and paper. Google it if you are interested in seeing it.
Some people understand instinctively which charts are better than others, but can’t necessarily explain why they know and picked the ones they did. Others, and no disrespect to many of my highly talented more technical colleagues, consistently pick awful ways to represent information. I have found that teaching business analysts, and others, to be better information designers isn’t actually that hard, once they understand the principles.
*(Image): Dr. Andrew Abela: Extreme Presentation Method