The Art of Visualization: A best practice guide by Iver van de Zand
Self-service business intelligence (BI) tools are a big hit and have permanently changed the BI landscape. The ease of use, performance, and agility towards ever changing data structures, while respecting data governance, are overwhelmingly appreciated by users.
With the capabilities of self-service tools to quickly create advanced analyses and visualizations, we touch our topic of the ‘art of visualization’. This art is about the interaction between the developer and the tool to create visualizations and dashboards that are crystal clear, understandable for the consumer, and can be interpreted without any doubt. Even more the art is about the ability to communicate essential messaging and data into actionable information.
Talking business intelligence to customers every single day, I am asked for my thoughts on the whole variety of ‘Christmas trees’ and ‘colorful paintings’ that are intended to act as core dashboards to monitor business insight. No doubt these dashboards are often pieces of art (graphical capabilites of the tools are really impressive), though they are questionable when it comes communicating data into actionable information. It takes an awful lot of time for the consumer to understand the dashboards and interprete them, if this can even be done. Is the developer to blame? In a sense, yes. However current tools make it so easy to use any graph and chart in almost any situation, clarity can quickly be created.
Time for guidelines on how to structure visualizations and dashboards:
Standards. Agreement on principles of visual development within your organization will greatly accelerate adoption and recognition of your dashboards and reports
Content. Avoid what can be avoided; less is more. The fewer graphs and charts needed to visualize your message, the better.
Multipage stories. Tools like SAP Lumira use stories that are multipage, allowing to spread graphs over several pages. This is preferred over a single page loaded with info.
Color coding. Only highlight or accentuate important data. Other colorcoding only distracts attention from your messaging.
Input controls. Inputs controls are a great way for end users to select and filter the data sections they are interested in. They take little space.
Data orientation. Use hierarchies and drill-able graphs but present at a summarized level. Present your data at the highest presentable level possible, and allow the users to drill where applicable. This keeps your dashboard clean and readable.
Legends. Avoid legends where possible. Separate legends next to charts and graphs use a lot of space. Instead use charts with embedded legends (see below example).
Gauges. I am asked every single week about gauges. Avoid them as they take way too much space. Bullet charts are a better alternative and offer more capabilites and better messaging.
Charts. Be careful with 3D and pie charts. Though they might look good, these charts take more time to interpret and are often unclear.
Tip: Numeric point graphs are very specific, take little space, and are crystal clear.
4ip: Radar or polar charts are very good when comparing facts over multiple dimensions.
Tip: Use heat maps for white space analyses
Tip: Use dynamic text. It makes text on a dashboard dependent to its underlying data. When used correctly, it often saves creating an additional chart or graph.
Luckily for us the ‘art of visualization’ attracts attention. The Web offers plenty of helpful guidance and information. Very interesting is the documentation from the Hichert IBCS group on notation standards in business communication that provides good ideas for standardization principles for visualization in your company. Happy reading !