Nicolette began her presentation with a quote from writer and entrepreneur Nancy Duarte that set the tone for the remainder of the night: “Data doesn’t speak for itself. It needs a storyteller.”
The importance of telling stories based on data lies in the stories’ potential to lead to positive change. Nicolette’s goals for attendees of her presentation were in line with this potential:
- Understanding how data storytelling helps cities improve
- Learning the steps to a basic workflow for finding, communicating and using an insight effectively in city operations and decision making
- Learning how to identify opportunities for applying a data storytelling workflow
In order to reach those goals, a data storyteller should follow seven specific steps. Nicolette used the example of making a decision regarding whether to close an underutilized library to explain these seven steps in a real-world context. Central Library, while currently underutilized, serves a critical function for a large Spanish-speaking population. Using this case study, Nicolette explicated the seven steps to data storytelling.
The first step is to frame the problem. This is usually done in the form of an explicit, actionable question that is relatively narrow in scope. Using the given example, this question is, “Should we close the underutilized library?”
The second step is to choose good data. To do this, it is necessary to determine what indicators will answer your question before gathering data. One can prioritize indicators by considering their relevance to the question, their actionability and the availability of the related data. Once data are selected, one can acquire and prepare it to be analyzed. In the example case, neighborhood demographics indicate the need for a library, and the chosen indicator is households that speak languages other than English at home.
The third step is to visualize the answers. The question being answered determines what visualization the storyteller should use. This is because different modes of visualization tell different stories. For example, bar or pie charts display composition, maps display differences over space and time series display differences over time. For the example case, a pie chart was appropriate in order to show the percentage of homes in which languages other than English are spoken in the neighborhood in which the library is located.
The fourth step is to record insights. The storyteller does this by writing down observations and findings in the visualized data. Then, the storyteller can combine observations from the data with contextual or qualitative knowledge. This knowledge could include things such as relevant events, local cultural or political context and subject matter expertise. For the example case, a data insight is that Central Library serves a large Spanish-speaking population. A further, contextual insight is that residents for whom English is a second language often have a hard time accessing information services in the Kansas City community.
The fifth step is telling a human story. This is a crucial step because people typically better understand a data-driven argument when it is framed within a human narrative. A narrative is made up of carefully chosen words, visualization that guides the eye and imagery that sets the tone. To design an effective narrative, one must start with the audience, define the message, highlight the best and prune the rest. It is important to keep in mind that good data stories get to the point, speak the decision makers’ language, speak to the place affected, use imagery and anecdotes to connect the data to real people and point to a recommended choice. For the library example, the pie chart is supplemented by an explanation of the critical function the library serves for the Spanish-speaking population in the neighborhood. This function is one this population is unlikely to find elsewhere.
The sixth step is to share the story with stakeholders. This is best done by making the results consumable by multiple audiences, regardless of their level of familiarity with data. How the storyteller shares will depend on the audience and goals. Possible methods of sharing include sending results directly to a decision maker, presenting at a community meeting and posting results to a social media platform. For the library example, the results could be shared directly with a committee in a small meeting context.
The final step is to take data-driven action. A successful data-driven story will lead decision makers to incorporate the story in order to improve policies, programs, perceptions and, ultimately, people’s lives. Good data storytelling will support these better decisions by crafting an informed and clearly communicated argument. This helps to make people-centered but data-supported suggestions for best practices moving forward in the context of a given problem. Regarding the library example, the data storytelling suggests that the library should not be closed.
If you are interested in learning more about data storytelling, Nicolette suggests the following:
- “DataStory: Explain Data & Inspire Action Through Story” by Nancy Duarte
- “Data Points: Visualization that Means Something” by Nathan Yau
- “Data Science & the Art of Persuasion” by Scott Berinato
Learn more about mySidewalk’s work here.
Code for KC hosts Learn Nights every third Monday of the month from 6-8 pm.