top of page

How to Tell Stories with Data

Normally on our blog, we discuss Laserfiche and its features and functions. However﹘in this case﹘ I wanted to use this post to discuss storytelling with data using Edward Tufte and Jim Stikeleather’s methods. At CPS we create a yearly audit for their Laserfiche system for our VIP clients. If you are interested in learning more please let us know. In the meantime, here are my thoughts on using data to create narratives that will benefit your organization.

Described as the “Galileo of Graphics” polymath Edward Tufte is both a statistician, artist, and the father of modern data visualization. Tufte’s work is why we can tell stories with data. Harvard Business Review contributor, Jim Stikeleather, proclaims, “visualization in its educational or confirmational role is a dynamic form of persuasion.” Hence, because of its confirmational, educational, and influential aspects, data visualization is perfect for persuasive storytelling for pitching business-driven IT efforts.

Storytelling helps the audience gain insight from the data presented. In particular when you wish to communicate to non-technical or non-analytical people. However, how does a person who has access to data and analytics find a story that the data supports? Stikeleather suggests using strategies derived from journalistic practice. In this spirit, here are some strategies to tell a story using your data and reporting:

  • Understand what questions you are trying to answer-Those of you who took a journalism class will recognize this approach. Ask yourself if this is a what, why, or how story? What stories tell what happened. These are typically the most like traditional journalism. Why stories dig at the underlying circumstances that caused the outcome. How stories, generally translate to How to address the problem. How stories examine various ways to improve the situation.

  • Know your audience- The most compelling narratives are written with the audience in mind. Within the broad term, ‘audience’ are multiple archetypes. The first question to ask yourself is, what does the audience know about the topic? What is their role? The narrative needs to frame what the audience already knows or doesn’t know. Here are some audience archetypes:

  1. Complete Novice- This type has had no exposure to the topic. They will need a higher-level overview, followed by a deeper dive.

  2. Generalist- The generalist has a broad knowledge of the topic. They are looking for an overview review and significant themes.

  3. Stakeholder/Expert- The stakeholder requires exploration, discovery, and options in great detail.

  4. Manager- The manager looks for in-depth, actionable understanding. Relationships and detail are essential. They are also looking for recommendations.

  5. V or C-level Executive- The executive needs to understand the overall significance of the data and conclusions weighted by probability.

  • Build context- In the background-during development, the best stories have meaning and a place in the workday of the audience or the audience’s clients. The possible impact of the story on the various audience members (and the action they need to take) has been considered. Journalists consider creating a context for the narrative part of building a relationship with the audience. In short, context is why the narrative, analytics, and data matter.

  • Be objective- Ideally, a set of graphics should not be biased. Tufte argues that “visual representations of data must tell the truth.” Although many graphics are developed to support a point, the graphic should be based upon what the data says, not what you would like it to say. Tufte proclaims that most charts are misleading. To this end, he developed a lie-factor calculation. The lie-factor is equivalent to the size of the effect depicted in the graphic, divided by the size of the impact in the data. For example, a number that is four times bigger than another will be perceived as sixteen times bigger if shown in 3D. Some other ways to reinforce objectivity are: clear labeling, graphic dimensions must match data dimensions, use industry-standard or recognizable units, and design elements should be subtle, so as not to compromise the data. Effects such as clustering, variable scaling, and alternative color palettes help inform the readability of the diagram. Decision-makers are particularly skilled at noting inconsistencies, which in term causes a loss of credibility.

  • Don’t censor- Unless you are utterly confident in interpreting the data, including more data in your narrative than you exclude — working out in advance how you work with outlier values and use discrete values when the data is continuous. Too much exclusion may cause your audience to lose trust.

  • Editing is everything- There’s an old truism that said a good piece should spend more time in the editing stage than it does in the initial creation stage.

bottom of page