The folks at Tableau have a nice little white paper 10 Tips to Create Useful & Beautiful Visualizations (registration required). Many of these are pretty straightforward, if still valuable reminders of best practice. I particularly liked Tip #1 – “tell a story” – as this is obviously critical for a powerful visualization. But I have one more tip – Tip 0 if you like:
Understand the decision you are trying to influence
At the end of the day the reason you present a visualization is to help someone make a decision. If the decision is a one-off decision then perhaps you are producing the visualization to push them in a particular direction. If, on the other had, the decision is one that must be made regularly (which sales region to focus on, which call center team to give extra training to, which cases to focus on) then the visualization should be designed to help them make a good decision, and probably to do so quickly.
To build effective visualizations, then, you will need to know:
- What decision you are trying to influence
- Who makes that decision
- What a good decision looks like (which KPIs are you trying to move and in which direction)
- How that decision is going to be made (what else besides the visualization is going to influence or constrain it)
In my experience far too few BI/visualization teams have an understanding of the decision-making they are trying to influence. They need to ask about decisions and for all but the most straightforward they need to model the decision so that there is agreement on how it should be made and on the role of the visualization/data being presented. While there are not yet widely used approaches for this, we have had some good success using decision requirements modeling (based on an emerging standard) to define the decision-making requirements of an analytic project and you can download our new paper on Analytic Success with Decision Requirements Models to read more about this.