Data Topics

Advertisement

Best Practices in Data Storytelling

By Jelani Harper on

Data storytelling is a critical aspect of analytics that was not given much attention until recently. An Enterprise Data World 2015 session entitled “Visualize This: The Art of Data Storytelling” featuring Kimberly Nevala and Bree Baich of SAS indicates that data storytelling plays an integral role in:

  • Advanced Analytics: The more advanced the nature of analytics used, the more data storytelling is required to simplify sophisticated predictive and Machine Learning algorithms and their results. Nevala noted: “The methods that we’re using to generate data and our insights are getting more abstract. As we get more abstract we have to think about effective ways to communicate what’s happening to a broad audience.”
  • Data Science: Data storytelling can be one of the most efficacious means for Data Scientists to translate the results of their analytics and algorithmic endeavors to those who are unacquainted with statistical principles.
  • Context: The crux of data storytelling is that it provides pivotal context for data and analytic insights that gives them meaning and makes them influential in many ways for the enterprise.
  • Deriving action: The underlying utility of data storytelling is that through an artful combination of images (provided by data visualizations), analytic insights, and words, it can produce a degree of cogency which makes action seem imperative while validating advanced analytics, Data Science, and their underlying business value.

Cognitive Origins

The practicality of data storytelling has largely emerged within the wake of data visualizations (which are related to, yet not synonymous with, storytelling) and a host of relatively new tools for providing critical data-driven narratives. Yet one aspect of this practice that Nevala and Baich emphasized was the advantages that utilizing a combination of media involving data, explanations, and, most importantly, images provides for basic human understanding and cognition. Nevala observed:

“Just like visualization plays on our natural abilities around pattern recognition, data storytelling plays on our brain’s architecture which is really focused on contextual or relational associations to aid memory. The fact is that stories stick and numbers don’t.”

Contextual Information

The primary way to get numbers to stick is by adding the sort of context that data storytelling innately provides. The human aspect of storytelling is implicit (and likely underserved) in analytics and involves explaining relationships between data sets and business objectives. It frequently requires a summary of the current realities of an organization or particulars of a business problem, and both shows and tells how analytics is indicative of a clear solution to such a problem. Additional contextual factors that story telling provides is based on the audience and how much it knows about analytics, data sets, business problems, and collaborative responses. Nevala commented that:

“This is where data storytelling has come into play because data storytelling is here to provide the context. It’s here to draw the relationship and walk us through a narrative that connects those different facets of it. What we’re trying to do here is move people from, ‘yeah I understand it’ to ‘now I know why this is important and I actually now believe what you’re telling me’.”


Tools

The images that typify a number of data discovery tools are crucial to this contextual evidence. As specifically related to storytelling, those tools include:

  • Visualizations: Available in any assortment of colors and patterns, visualizations can provide graphical representations of data that provide visual points of comparisons between data sets. Nevala explained that: “What visualizations do not do well in and of themselves is make a compelling argument. They might help you understand, they may make you believe, but they certainly do not make you ask [questions] without some additional context.”
  • Annotations: These tools enable users to verbally “mark-up” reports and dashboards, augmenting visual information with conventional verbal information, so that more than one storytelling medium is used to convey the significance of data.
  • Storyboards: Storyboard capabilities are used to create narratives that can guide users through visualizations, providing the context and meaning of the data.
  • Infographics: Infographics provide key design elements so that users can visually represent data.

Data Science

Data storytelling and its various tools have the potential to be as valuable for Data Scientists as other facets of analytics such as open source R, sandboxes, and data lakes. Its primary value for these professionals lies in the fact that by incorporating storytelling, they are no longer simply quantitatively inclined, glorified statisticians who care more about their precious numbers than they do human interaction with daily business issues. Instead, they become gifted storytellers with a conviction that might otherwise elude their statistical analyses. “It’s given us a common lexicon,” Nevala said about data storytelling, “that new ontology that we can talk—on like terms—to each other.” In this respect, storytelling is a key means of enabling the statistically competent to interact with those that might be otherwise, while improving both of their jobs in the process.

Beyond Data Science

Although the drivers for storytelling are intrinsically related to the need for analytics and Business Intelligence to inform decision making, Nevala indicated that the most prominent one is even more profound. The more that analytics is used to refine and enhance business processes, the greater propensity those involved in those processes will learn that some of the core beliefs about them and their organization are either unfounded or function differently than was previously perceived. In this respect, storytelling not only influences action based on analytics, but can actually help to reform and improve core business processes. Nevala mentioned that, “The need to really communicate and directly manage that level of change is directly proportional to the need for data storytelling, and storytelling in general.”

Implementation

Despite the fact that many of the specifics for data storytelling hinge on the particular use case, intention, and business problem being solved, there are some general principles (which bear some similarity to writing a data focused article) that constitute best practices and are detailed below in sequential order:

  • Set-up: The initial moments of a data storytelling presentation should begin with an intriguing opener to foster curiosity. The time-honored example is by illustrating a business problem or its effects on the organization.
  • Context: It is necessary to illustrate the current realities that pertain to the data or the business problem, which requires storytellers to select visualizations, determine the intention of the action they hope to produce, and identify tools and other logistical information.
  • The Options: This step represents the point in the story in which the presenter persuades listeners of the utility of his or her proposal based on analytics. It directly involves the data’s implications and, according to Baich, is best represented visually with contrast in which the “risk and intrigue” of a proposition is denoted. Contingent concerns involve data available for use and any governance processes that may be involved.
  • Call to Action: The summation of the storytelling presentation, the call to action emphasizes what exactly the company can do—as readily as possible—to address the issue or business problem by providing a solution.

Resolving Analytics

With the plethora of self-service, SOA, and data discovery options in the contemporary world of Data Management, facilitating analytics to glean insight from data has never been easier. The core problem that has persisted with analytics is creating action from that insight—which pertains to a whole host of factors including Data Quality, Data Governance, timeliness, and more. As Nevala observed, “Understanding is not necessarily enough. How do you make me believe in that [data] and fundamentally start to act on that information? That’s been the fundamental conflict with analytics for a long time.” Data storytelling is designed to illustrate the path to action while incorporating as much reason and what Baich refers to as “evidence-based data” to denote the proper course of action. Nevala asserted that, “Story grounded in fact is a principle and a guideline for how we want to behave and act. In that way we can start to encourage better behavior.”

Leave a Reply