Data storytelling
BUS5AP Analytics in Practice
Stu Black
Agenda
• Welcome
• Guest Speaker – Alon Ellis
• Assignment 1 Marks – a Stakeholder Expectations Learning Opportunity
• Data Storytelling
• For Next Week
• Open Q&A
• Close
Agenda
• Welcome
• Guest Speaker – Alon Ellis
• Assignment 1 Marks – a Stakeholder Expectations Learning Opportunity
• Data Storytelling
• For Next Week
• Open Q&A
• Close
Agenda
• Welcome
• Guest Speaker – Alon Ellis
• Assignment 1 Marks – a Stakeholder Expectations Learning Opportunity
• Data Storytelling
• For Next Week
• Open Q&A
• Close
Expectation Mgmt… a Learning Opportunity• All Assignment 1 papers have been marked… hopefully moderation will have been completed and you
should have your marks accessible
• A large part of this subject is about expectation management‐ Setting expectations through articulating a proposal‐ Delivering against that expectation‐ Learning the gotchas so that you can avoid them on your journey
• You have a rare advantage in this course… I tell you my expectations in advance… I publish the marking rubric
• My suggestion… if you got less than 27 / 30 marks… undertake a self-assessment‐ Review the rubric… read it carefully‐ Assess your own paper against that rubric. (Please be sufficiently self-critical)… what would your
mark, if you marked your paper against the published rubric ‐ Review the marks (and the comments) provided in the LMS and granularly compare against your
self-assessment. Identify specific deviations and/or broad trends‐ Develop a view as to what has driven the difference in expectations, and determine what you will do
in future assignments to close that expectation gap
Agenda
• Welcome
• Guest Speaker – Alon Ellis
• Assignment 1 Marks – a Stakeholder Expectations Learning Opportunity
• Data Storytelling
• For Next Week
• Open Q&A
• Close
Why story telling?
• From your selected BadViz…
• How effective was it?
• What would Tufte say about this?
Black’s Analogue to the Theory of Information
If the receiver can determine the
next bit of a bitstream, then the
message carries no information
If you present a visual to a decision maker, and that decision maker
takes no action, did you create any value?
“Insights” or Action-Inducing Insights
If you present a
visual to a decision
maker, and that
decision maker takes
no action, did you
create any value?
• How many of you said (in Assignment 1) that
you would deliver:
‐ “Insights”?
‐ “Visualisations”?
‐ “Descriptive Analytics”?
• Did you articulate the potential actions that
would be taken once the insights were
delivered?
A real-world example – SMSF holdings
What action should I take?
SMSF worked through
Why story telling?
• Humans are wired for story
• A successful story produce and neutralise anxiety
• Not just data scientists, most professions need story telling skills
• Examples
• “Families are doing it tough” but a liberal government will reduce your cost of
living by [you can put anything here literally]
• “Our attrition rate has been increasing YoY and above national average” but
we can [develop an intervention program to improve retention]
What is data storytelling?
• Example: describe a corgi to someone
• Is it easier with numerical raw data, or simply a photograph?
• The same applies to data
• A visual (e.g., graph) is usually better than showing the raw numbers
• Creating a graph is not data storytelling
• Data storytelling requires combination of data, graphs, key observations and
conclusions that are linked through a narrative
Data storytelling
• A good narrative enables one to go beyond the relay of information
• It becomes a powerful mechanism of persuasion
• Gets a point across to the audience quickly with aid of data
• Improves retention of the key information that you want audience to
remember (up to 22x)
• BUS5VA taught you some good practices
Some good tips
• Use the right data
• Is it out of date?
• Is it open to interpretation, or questioning?
• Is the data suitable for the intent?
• Synthesise
• Use combination of data, or focus on individual parts to drive a point
• Example: contrast current wage growth over past wage growth in your
discussion of housing affordability
Some good tips (cont’d)
• Make it personal and real
• How does a lawyer convince the jury?
• Example: discuss housing affordability in Melbourne
• Don’t overload your narrative
• Learn from Hollywood movies
• Never try to say too much
Story structure
• Start with the three “knows”
• Know what you want to say
• Know what your data is saying (and if it supports the first “know”)
• Know what you audience want to hear
• Sketch out your story
• Stick to good data storytelling genres
Story telling genres
• Linear logic
• Start at the beginning and move
linearly to the conclusion
• Reverse of the above
• Change over time
• Linear but time-driven
• Flow diagrams and road maps
Story telling genres
• Linear logic
• Start at the beginning and move
linearly to the conclusion
• Reverse of the above
• Change over time
• Linear but time-driven
• Flow diagrams and road maps
Story telling genres
• Linear logic
• Start at the beginning and move
linearly to the conclusion
• Reverse of the above
• Change over time
• Linear but time-driven
• Flow diagrams and road maps
Story telling genres
• Compare and contrast
• But make sure you are comparing
apples to apples
• Progressive depth
• News are usually written this way
• Personalisation
• People are interested in things
related to them
Story telling genres
• Compare and contrast
• But make sure you are comparing
apples to apples
• Progressive depth
• News are usually written this way
• Personalisation
• People are interested in things
related to them
Story telling genres
• Compare and contrast
• But make sure you are comparing
apples to apples
• Progressive depth
• News are usually written this way
• Personalisation
• People are interested in things
related to them
Good storytelling doesn’t require
the construction of complex
visualisations
Often, the persuasion we chase
doesn’t warrant the investment
or time.
Better data storytelling with good visuals
• Many large organisations develop their own “best practices” that guide you to
making good visuals
• Most of the time, they are guides rather than hard rules
• Some good starting points
• Create focal points
• Eliminate clutter and confusion
• Have a takeaway message, or call to action
• Think different
Create a focal point
Eliminate clutter and confusion
Have a takeaway message
Click image to open solution templates
Think different
Exercise
Exercise
Exercise
Exercise
• What is the graphic seeking to
communicate?
• What is right with this
visualisation, and what is wrong?
• What should be done differently?
Exercise
And there’s the usual best practices
• Less is more
• More is more (sometimes)
• Keep it simple
• Sometimes, fancy is just that -
fancy
Summary
• Good analytics need good data storytelling to go with it to be successful
• A few good principles to follow
‐ Start with the three “knows”
‐ Think of the structure of your data storytelling
‐ Think of the action that you want the reader to take
‐ Identify relevant best practice
‐ Develop visuals, using those identified best practices
• Keep practicing
Agenda
• Welcome
• Guest Speaker – Alon Ellis
• Assignment 1 Marks – a Stakeholder Expectations Learning Opportunity
• Data Storytelling
• For Next Week
• Open Q&A
• Close
For Next Week…
• Research and Identify at least one Project Management “Fails”
• Identify the following
‐ What Project management Approach did they use?
‐ What were the challenges that project faced?
‐ What could they have done differently to avoid the ‘fail”?
Agenda
• Welcome
• Guest Speaker – Alon Ellis
• Assignment 1 Marks – a Stakeholder Expectations Learning Opportunity
• Data Storytelling
• For Next Week
• Open Q&A
• Close
Agenda
• Welcome
• Guest Speaker – Alon Ellis
• Assignment 1 Marks – a Stakeholder Expectations Learning Opportunity
• Data Storytelling
• For Next Week
• Open Q&A
• Close