While the concept of data storytelling has been gaining popularity, I have noticed it hasn’t always been used correctly or consistently. As I’ve worked in the fields of analytics and business intelligence, I’ve seen the term used haphazardly by both data experts and product vendors. Unfortunately, it has left many people confused about what data stories are and how to best create and deliver them.
When you add in the excitement about technological advances such as ChatGPT, it’s only going to get messier and more confusing if people don’t have a fundamental understanding of what data stories are and how they’re created.
A major part of the problem is we lump three related but distinct steps together under the broader label of data storytelling. To clarify and better understand what data storytelling is, I find it helpful to take a closer look at two upstream steps that must occur before a data story is even told: storyframing and storyforming.
I developed the concept of storyframing when I joined a business intelligence SaaS provider. As I was preparing to deliver a breakout session on data storytelling at its upcoming customer conference, I had to rationalize the role of dashboards within the context of storytelling. I knew my new employer expected me to highlight the platform’s “data storytelling” functionality in my session.
However, there was a major problem—I’ve never believed dashboards tell data stories. Nothing I saw in this BI platform (or any other) has convinced me otherwise. Because the data in automated dashboards is constantly refreshing, potential story points become moving targets—coming and going, shifting and evolving. In addition, you can’t predict where a “story” will emerge from the data—it might be related to one key metric one time and a completely different one the next time.
The fundamental challenge is most dashboards are designed for reporting and monitoring the business. They are exploratory tools—not explanatory ones. Furthermore, when it comes to telling the story, storytelling is more than just a layout option that enables users to add comments or generative AI that produces descriptive text to accompany the charts.
In my view, data stories are built on insights, not just information. Most organizations have an abundance of data and information, but insights are much more elusive. When not all your data is equally essential, it’s important to know which metrics and dimensions are critical to the business. Before you start any analysis to find insights, you must first have a good grasp of your audience’s needs, interests, and key questions. As an analyst and data storyteller, the better you comprehend what’s essential to your audience, the better you can focus or “frame” your search for meaningful insights.
It’s important to clarify you are not framing the telling of a specific story. At this exploratory stage, you don’t yet have a narrative to share. Instead, you are framing the discovery of potential insights and their subsequent data stories. Essentially, rather than looking at all your data, you’re prioritizing where the focus of attention and analysis should be. Look here and here, not over there. By establishing the key metrics and dimensions during the storyframing step, you ensure the data discovery process is both targeted and more productive.
While an automated dashboard may not be the best tool for data storytelling, it is useful during this storyframing step to quickly generate useful observations. A well-designed dashboard should contain information that is aligned with the audience’s key business objectives, creating an invaluable window into business performance. As the dashboard updates with new data, it can highlight results that are unusual or unique. Even though they may not tell data stories, dashboards can help people easily spot key anomalies, trends, and patterns in the data (make observations), which can lead to insights with further analysis.
Once you have a potentially promising observation—an unusual anomaly or interesting trend—you need to understand ‘why’ it occurred. Identifying weekly sales increased by 200% won’t mean much until you determine why the bump happened. On their own, observations aren’t actionable. For example, in the case of the 200% increase in weekly sales, you will want to first identify the potential root cause before deciding how to respond to it:
If you find the root cause was plausible and the resulting effect was expected, you have an explanation—not necessarily an insight. For example, you may discover the marketing team launched a product discount that coincides with the 200% increase in sales. In this case, on the surface, this explanation makes sense and isn't an insight. As psychologist and author Gary Klein stated, an insight represents “an unexpected shift in the way we understand things.” If the cause and its effect are surprising, you may have found an insight that could benefit your organization. Without an insight, you’re missing the central ingredient of a good data story.
Insights don’t magically fall into your lap—despite what analytics vendors keep telling us. Most analytics tools are designed to help us find ‘what’ happened (descriptive analysis) but can rarely explain ‘why’ something happened (diagnostic analysis). Unearthing insights can take significant time and effort to uncover. It is during this more involved discovery phase that a narrative can begin to form and take shape inside your mind as the analyst.
The storyforming process spans both the data analysis and data interpretation. Data analysis is the process of examining something in more detail to better understand it. Data interpretation is about making sense or meaning of these findings and drawing conclusions. It’s the combination of these two separate but related activities that can lead to new insights. While the quality of your analysis will be impacted by your analytical skills, the strength of the data interpretation will be dependent on your domain or business knowledge.
Frequently, these two steps are often presented as sequential stages—you complete all the analysis and then move on to interpreting the findings. However, in my experience, they are interwoven and iterative. As you uncover findings in the data, you can draw conclusions based on what it represents, which will then inform the focus or direction of your data exploration. In the following diagram, you can see the back-and-forth nature of how you analyze and interpret the data that can ultimately help identify an insight.
Through this process of analyzing and interpreting the data, you start to assemble the pieces of a potential narrative for a data story. If you’re fortunate enough to uncover a meaningful insight (not all analyses necessarily do), the findings and conclusions made during the analysis process will have formed a rough story in your mind.
At this point, you must now decide how you will proceed with your insight. If it would be valuable to share with an audience, that’s when data storytelling enters the picture. You must further refine the story that developed in your mind during the discovery process and communicate it in a way that helps others form a similar narrative in their heads and incite them to take action.
A good data story will guide your audience through the numbers, mimicking the storyforming process that you experienced but in a streamlined, guided fashion. Essentially, the audience will perform an abbreviated version of your data analysis and interpretation as they listen to your data story, identifying the same findings and drawing similar conclusions to yours.
Part of the challenge of data storytelling is recognizing even though the numbers spoke to you, they may not speak the same way to your audience. Mastering the art and science of data storytelling ensures your insights are understood, appreciated, and will drive change. Effective data storytelling occurs when your audience walks away from your data story with the same insight(s) that you discovered in your analysis. In addition, they are better positioned to understand the problem/opportunity and are motivated to act on your insight(s).
In summary, you can see the differences between each stage by the unique questions that each stage seeks to answer.
Both the preceding steps of the storyframing and storyforming play an influential role in the data stories you craft and tell. By better understanding how they influence and develop the finished product (a data story), you can better grasp what data storytelling’s real purpose is in the analysis process—to explain insights in meaningful ways so they inspire action and change.
Note: I first came across the concept of “storyforming” in an article by analytics consultant, Meagan Longoria. She referenced a conversation she had with data visualization expert, Andy Kirk, who introduced the term to her. However, what they refer to as storyforming is what I call storyframing. A dashboard frames the potential stories and then when someone analyzes and interprets the data, they begin to form the story.
Effective Data Storytelling teaches you how to communicate insights that influence decisions, inspire action, and drive change.