UXPA – Visualizing Data for People

consThe event featured a very interesting speaker:

Paul Derby is a Senior Experience Designer within the Honeywell User Experience design studio. Paul has a PhD in experimental psychology (human factors) from Texas Tech University. At Honeywell, Paul focuses on UX research and design within the process industry (e.g., oil/gas, petrochemical, etc.). Currently, Paul is leading multiple UX efforts to improve data visualization products within this domain.

Paul spends time talking with engineers in refineries to understand their needs and works with designers to assure that the  visual monitoring dashboards provided by the company fit the need of the end users.  Product team includes people from different disciplines: marketing , engineers, designers, UX specialists, etc.


Paul experienced a significant transformation at Honeywell: the company now has a VP of experience design, over 25 designers throughout the organization, and plenty of opportunity for UX work.  It is interesting to note that Paul is a psychologist, rather than a designer or an engineer, what makes a complete sense 🙂

Ha ha – I remember attending my first university (majoring in psychology), and during that time “industrial psychology” class did not seem appealing area of specialization for most of my class.  A couple of decades later the field seems to gain significant popularity 🙂

So….  what is a dashboard?


Dashboard design principles based on human physiology and psychology:

Designing to support how we see color:

  • Green, yellow and red should be reserved for important information
  • Use color as a redundant backup – design visualizations that work in monochrome to accommodate color blindness

Designing to support attention:

  • Redundant coding – use color/shape/motion to draw attention
  • Avoid clutter (avoid meaningless pictures and effects
  • Support visual scanning through structure – use Gestalt principles to group objects for easy scanning

Designing to support working memory

  • Place related information in close proximity
  • Avoid interpretation – express important data directly and visually (using a graph to express data that is above ore below certain point
  • Avoid excessive details (5,23742 rather than 5.2)

Good chart to support working memory – the information is easy to perceive at a glance


Design to support situation awareness

  • Single screen
  • Context (use spark lines)
  • Leading indicators (display what is likely to happen rather than what already happened) – use colors to indicate that the numbers are approaching a certain stage, etc.  Example below highlights “situation awareness” principle applied to three numbers:
    • Oranges sold – 32
    • Bananas sold – 45
    • apples lost      – 98


Below are examples of bad dashboards (what seem to be relatively easy to find).  However, if asked during the dashboard development process, many users will prefer “pretty,” more colorful dashboard, even if it won’t be as easy to use in the day to day basis for the goal of process monitoring…

If you ask the users, they will want something interesting to look at…



Why the dashboard is bad: difficult to read (black background), unnecessary graphical elements, too much space taken by a simulated gadget showing one number, etc.



Why this dashboard is bad: not clear which information is important, too busy; overlooked Gestalt principles for grouping, etc.



Why this “pretty” dashboard is bad: the dashboard is trying to replicate live equipment (adds confusion), colors used randomly rather than communicate important information.


Good dashboard


Why this dashboard is good: the process is represented symbolically in the background, colors are used to convey meaningful information; redundant coding for color and shape


bookTwo books were recommended at the end of the presentation


Paul mentioned an organization created by process industry companies and educational institutions  to prevent “abnormal situations” in the industry.