Knowing how to use Tableau doesn’t mean you'll be able to design effective dashboards. If you want to create dashboards that deliver valuable insight, perform well, and have visual impact, you'll need to apply Data Visualization Best Practices.
In this webinar, you'll learn the science behind Data Visualization Best Practices. Cognitive psychology helps us understand how the human brain perceives information in a dashboard, and we'll teach you how to use this knowledge to optimize your designs.
7. According to Stephen Few, one of the original data visualization
evangelists:
A visual is a visual display of the most important
information needed to achieve one or more
objectives; consolidated and arranged on a single
screen so the information can be monitored at a
glance.
8. Not all users are the same:
◦ Not all users have the same level of analytical expertise.
◦ Not all users have the same familiarity with the subject matter.
◦ Not all users answer questions from the same perspective.
It is highly unlikely, or desired, to meet these disparate needs with a single
visual.
9. There are three general types of analytical visuals that address end user
needs:
◦ Strategic/Executive – A high level view of a question or line of inquiry
◦ Operational – A regularly updated answer to a question or line of inquiry
◦ Analytical – Interactive view providing mechanisms for a variety of investigations on a
specific topic.
Knowing the type of visual you’re creating prior to even starting will help
you focus the design efforts as you get started.
11. The study of the neural processes in the brain underlying visual perceptive
and cognitive functions.
12. What is the difference between perception and cognition?
◦ Perception is the organization, identification, and interpretation of stimuli from your
senses. An overly simplified way to describe is acquiring information through the
senses.
◦ Cognition is the processing of information and acquiring knowledge through reason,
intuition, and perception. A simpler way to think of it is cognition is the processing of
perception into knowledge.
Perception is fast, cognition is much slower. Is there something we can do
with this information?
13. How is this relevant to us?
◦ We can make use of visual perception principles and design visuals that result in an
immediate understanding of the information that is arising from the data.
Visuals that properly implement the findings of visual and cognitive
neuroscience will result in “intuitive” visuals.
The recommendations derived from visual and cognitive neuroscience are
referred to as “visual best practices” or VBP
Let’s look at a couple of examples… in each example one visual properly
implements VBP, while the other does not. Let’s see if you can guess which
is which.
17. Seeing (perception) and comprehending (cognition) occur all the time,
every day from second to second.
The shorter the time between “seeing” and “comprehending”, the more
productive your end users will be and the more “intuitive” your
visualization will be.
You should apply scientific findings when designing your visuals.
Above all else, understand what works in what situations, what does not,
and the difference between the two.
18. The human brain can process small amounts of visual information very
quickly, but only for a very short period. This is called “Visual Short-Term
Memory” (VSTM).
Each element in the visual must be “seen” (perception) within roughly 40
milliseconds… any longer and it must be processed “attentively”
(cognition). If you take advantage of VSTM your end-users can interpret
entire visuals within a few seconds.
To make use of VSTM requires us to know what works and what does not
work when visual processing is occurring. Things that work well with VSTM
are called “pre-attentive visual features”.
If you do not make use of VSTM the brain must send the visual information
to another area of the brain for more robust processing, but this additional
processing is considerably slower than VSTM.
19. How many 5s are there?
Answering this question must be done consciously, or “attentively”.
◦ We can process it attentively either using linguistic skills (find the number 5s), or
◦ We can process it attentively using pattern matching skills (find things shaped like ‘5’)
Either of these options require “attentive” processing.
20. Try again… How many 5s are there?
Answering the question now can be done subconsciously, or “pre-attentively”.
We took advantage of a VSTM pre-attentive feature called “color hue”.
This technique is known as “perceptual popout” and is often implemented as a
“highlight” in visualization tools.
21. Have our sales increased or decreased over the four years?
Which category has grown the fastest?
This cannot be processed pre-attentively as text and math require post-
attentive processing.
22. Have our sales increased or decreased over the four years?
Which category has grown the fastest?
This can be processed pre-attentively.
40. Categorical Data – Data that logically belongs to a group based on
characteristics.
◦ North America, Europe, Asia
◦ Eric, Sheri, John
◦ Pizza, Twinkies, potato chips
Ordinal Data – Data that logically belongs to a group based on
characteristics, but also has a logical sequence to group members:
◦ Gold, Silver, Bronze
◦ Doctoral, Masters, Bachelors
◦ May, June, July
41. Quantitative Data – Data that defines “how much” of something there is.
◦ Sales: $10,000; $1 million; $4 billion
◦ SAT scores: 1100, 1500, 2400
◦ Fertility rate: 1.8, 2.4, 7.6
◦ Temperature: -26 ℃, 20 ℃, 46 ℃
42. After decades of scientific research the best combination of pre-attentive
approach for each type of data have been identified:
44. Follow a methodology
Define the central question
Know the audience
Determine “next steps” to support
Classify the deliverable
Profile data
Apply the most effective visual features
Design iteratively
45. Define a process by which you:
◦ Obtain design requirements
◦ Access source data
◦ Design visuals
◦ Release and promote
Having a methodology allows you to continuously improve and achieve
consistent quality in your deliverables.
Checklists and/or forms are very helpful.
Unilytics recommends our UD3 methodology for dashboard design and
construction.
46. What question should the dashboard answer?
Your audience will never clearly interpret your visual if you, as the
designer, didn’t clearly define the question to start with.
If you don’t know the question you’re answering, how can you answer it?
If the “question” you’re answering is a paragraph, you haven’t really
simplified and distilled your requirements… a sentence or two should be
adequate to express almost any data inquiry.
Corollary questions are OK, but keep visuals focused on that main
question. If you’re building a visual that is hard to trace back to the central
question… it shouldn’t be included.
47. Not everyone interprets or understands data the same way you do, or the
same other users of the visual will.
Are they technical or not? Do they understand statistics? How will they
access your visuals?
How will they understand KPIs when they look at them?
Will they know what to do after getting an answer from your visual?
48. Some dashboards present “information only”, while others support further
action on the part of your audience (“next steps”). Know which of these
you’re creating.
If your visual supports further action, know what information your
audience will need to perform these actions.
Summarize the core information for “next steps” as succinctly as possible
in as centralized a location as possible.
49. Remember the three main types of analytical visuals, and know which one
you’re building:
◦ Strategic/Executive – A regularly updated answer to a question or line of inquiry that
frequently monitors operational concerns in response to events or on an ad-hoc
basis.
◦ Operational – A high level view of a question or line of inquiry that is usually
answered in a routine, specific way and usually presents KPIs in a minimally
interactive way. d
◦ Analytical – A highly interactive view that provides a variety of investigative
approaches to a specific central topic with a few corollary contextual views.
50. Remember that data comes in three flavours:
◦ Categorical – Data that logically belongs together, such as: North America, Europe,
and Asia.
◦ Ordinal – Data that logically belongs together and has a logical sequence: gold, silver,
and bronze medals.
◦ Quantitative – Data that defines “how much” of something there is: $1 million in
sales, 20° Celsius, 150 defects.
51. Use the most effective visual feature for your type of data:
52. A good way to kill any business intelligence methodology is to force
designers to wait for 100% complete requirements.
The best way to design is to bring audience representatives or stakeholders
in during critical points and show them what is being created.
Let the audience representatives and stakeholders do “mid-stream course
corrections” as early as possible to ensure minimal lost or “redone” work.
Remember, you’re communicating visually… not linguistically. It’s very hard
to guarantee you’re “seeing” in your head what the audience is expecting
to “see” from their end… so check in with them routinely during design
and construction.
53. Follow a methodology
Define the central question
Know the audience
Determine “next steps” to support
Classify the deliverable
Profile data
Apply the most effective visual features
Design iteratively