Effective Data Storytelling and Visualization
As a self-styled cinephile, I trained myself to scrutinize a film from different aspects when watching it for the second or third time. Dialogues, acting, cinematography, these all help me judge a film for its meaning and craftsmanship, and increase the joy of viewing, but the most crucial part of this experience is still the story.
Freytag’s Pyramid in Storytelling
The idea of story structure beginning with Aristotle in his Poetics, but the full establishment is not until 1860s when German playwright Gustav Freytag developed a structure now known as “Freytag’s pyramid”. In his dramatic arc, Gustav highlighted seven parts that he considered as essential in storytelling:
- Exposition: The background of the story. The scene to convey information about characters, world, or history to the eavesdropping audience.
- Inciting Incident: My favorite explanation of inciting incident came from Robert McKee’s book Story, in where he summarized it as a trigger event that “radically upsets the balance of forces in the protagonist’s life”.
- Rising Action: The protagonist must react to inciting incident, and these reactions always lead to more complicated situations or problems.
- Climax: The turning point that places the protagonist under the maximum pressure of his life. Either through greatest tension between the protagonist and antagonist, or a choice between irreconcilable goods, the lesser of two evils, or both.
“This dilemma confronts the protagonist who, when face-to-face with the most powerful and focused forces of antagonism in his life, must make a decision to take one action or another in a last effort to achieve his Object of Desire.”
— Robert McKee, Story.
- Falling Action: “The hostility of the counter-party beats upon the soul of the hero”. The story shifts to the choice made by the protagonist, sometimes this could be a reversal.
- Resolution: The problem need to be resolved, and in ideal cases, “one final action by the protagonist settles everything”.
- Denouement (or Catastrophe): The story reaches its conclusion, but the climactic effects could continually spread…
One point I would like to complement Freytag’s Pyramid is that, the denouement need to have some sort of elevation from exposition, and that elevation is usually character development.
That “boy killed the dragon, and everything get back to normal”, ending never gonna work, unless the protagonist and the audience learned something from this journey.
When you slide stories into scenes, it’s not hard to realize that most of our great fictions, plays or films follow this structure. Although it worths to mention that these parts are not only separated by the development of the story, but also the development of protagonist’s character arc.
Too often, character and story are viewed as separate entities — to the point that people often compare them against each other, trying to determine which is more important. K.M. Weiland thinks about this relationship differently:
“Plot and character are integral to one another. Remove either one from the equation (or even just try to approach them as if they were independent of one another), and you risk creating a story that may have awesome parts, but which will not be an awesome whole.”
— K.M. Weiland, Creating Character Arcs: The Masterful Author’s Guide to Uniting Story Structure, Plot, and Character Development.
Freytag’s Pyramid in Data Storytelling
In data storytelling, without doubt that data is our character, and the insights & findings are the story. Without data, there is no story. Without story, data is boring and hard to consume by the audience. With a well-designed story structure, our data storytelling can be more intriguing. Until we have it, the data wait like suffering patients praying for a cure.
The first step of data storytelling is understanding the perspective of the audience. Who they are? What are they care about? What are the most important insights you want to rely? These questions need to be answered before crafting the story arc in order to make sure the right insights to be heard, and appropriate actions initiated. Then we can re-map Freytags Pyramid’s seven parts play in data storytelling context:
- Exposition: The background information: What’s the status quo? (either for a specific entity or across industry) and why is it important?
The exposition part acts like setting up a hook. It need to be either interesting enough or important enough or both to get a grip on audience.
Example: There is a clear correlation between site performance with business outcomes. Better performed site tends to generate more revenues… Here is the historical comparison between you and your competitor… - Inciting Incident: Data shows that something unexpected changed. What is that?
Example: While the overall site performance didn’t change a lot, your revenue declined by 10% compare to last quarter. There are seasonality reasons lie behind the scenes, but your competitors revenue only dropped about 5%. - Rising Action: A series of insights found through data analytics.
Example: We narrow down the site performance by page, and found the home page got slower. However, due to the improvements on product pages, this degrade wasn’t visible at overall level.
Further look into this, Desktop users don’t seem be impacted a lot, but the Mobile site got significantly slower with much higher bounce rate on browser X. The sales on browser X was reduced by a third… - Climax: The conclusions from the analysis, and what’s expected consequence if nothing changed.
Example: The home page is broken on browser X due to the failure of Y. If we don’t fix that, we would expect continuously customers and revenue loss. - Falling Action: What is the best immediate action for now. What are the options for long term solutions? Pro & Cons comparison.
- Resolution: The expected outcome after resolving the issue.
- Denouement: Further actions. How to prevent this from happening again…
Effective Data Visualization in Aid with Data Storytelling
Most of time, data visualization alone is not sufficient for delivering the analysis and insights, as the audience may or may not share same view or have comparable fluency with the data. But without visualizations, it’s almost impossible to have a good data story with strong impression and empathy.
There are some common mistakes in data visualization that could lead to ineffective storytelling:
As you can see, some simple tweaks around data, metrics and visualization types can tell a much stronger story.
