BOTH graphs are saying the same thing. Use one of them to answer the following question:Imagine you’re Sr. Sales Director evaluating the 5 different channels through which your company brings in revenue. Comparing 2 of the channels, how does the revenue for ‘Telebriz’ compare to the revenue for ‘Partner’?
Same thing here, 2 graphs showing the same information:Imagine you’re a CIO taking a look at your company’s profit margin indicators. What is more stable over time, our revenue or our operating costs?
Simple question: who is doing better, the East Coast or the West Coast?
Why dashboard design should be (but usually never is) based on cognitive science unconference
Why Dashboard Design
should be (but almost never is)
based on Cognitive Science
Catalog Telebriz Partner Internet Direct
Revenue by Sales Channels
Jan Feb Mar Apr May
(millions) Costs vs. RevenueA
$0 $50,000 $100,000 $150,000
Revenue by Region
• Most dashboards are far less effective than they
• Is it the responsibility of the UX community to fix
• It is a difficult problem to solve.
– Proper graph construction is not taught in school
– Data visualization of business intelligence is an
extremely small niche area, even within UX!
– There are usually many organizational and
institutional obstacles to doing things the right way.
Why is this important?
• Business intelligence is critical to the
operation of virtually all modern institutions
• Visualization is the perhaps best way to
• Dashboards are a common tool; that tool is
often broken; and it needs to be fixed.
What approach she we take?
• Two opposing approaches to making graphs
– Thinking like an artist?
• Striving to express oneself, using the data
– Thinking like a translator?
• Striving to translate the data from one language to
– The mathematical language of the data
– The language of the human sensory, perceptual and cognitive
Conclusion from the 3 Demonstrations
– Color attributes can help performance
– Color attributes can hurt performance
– Humans are good at judging length, and bad at
judging area of objects
– These perceptual phenomena are universal (so why
don’t we use them!)
• Know the language of the human
sensory, perceptual and visual systems
“Language” for Human Quantitative Judgment
Category Attribute Quantitative
Color Hue No
Intensity Yes, but limited
Form 2-D position Yes
Line length Yes
Line width Yes, but limited
Size Yes, but limited
Intensity Yes, but limited
Motion Flicker Yes, based on speed but limited
What does this mean for us Uxers?
• The UX practitioner community should stay informed
on the best practices for data visualization.
• Perhaps we should aim to steer data visualization
design efforts in our own organizations
• Perhaps we should also aim to be the thought leaders
in this area as the field continues to develop over time.
• Please take the handout sheet (up front), which is a basic
guide on how to choose appropriate graphs for effective data
• Most useful & reputable authors in the field:
– Edward Tufte
– William Cleveland
– Stephen Few
Sparse data Rich data