7 data storytelling examples: How to turn data into stories

1. Harness 'data communications' for effective data storytelling

The first data storytelling style comes from Kim Herrington, a senior analyst at Forrester Research. Her variety of storytelling is fueled by a variation on internal corporate communications: data communications, or DataComs, in Forrester's parlance. The DataComs concept is aimed at helping teams of so-called insight professionals optimize their communication and collaboration processes to better inform business decision-making.

As you might expect with a consulting firm, Forrester has a formal definition of data storytelling: It's "the art of defining, analyzing, structuring, rehearsing, condensing and communicating insights found within data sets." Data literacy, storytelling's usual companion in analytics initiatives, is "the ability to recognize, evaluate, work with, communicate and apply data in the context of business priorities and outcomes." DataComs is designed to fill the gap between literacy and storytelling for more effective communication of data insights -- for example, through internal committees and user communities.

From a practical data storytelling standpoint, Herrington emphasizes preparation. For example, she begins by defining how she wants her audience to think differently about business issues based on the data insights. Then, what does she hope they'll do differently as a result? That helps frame the data story and its conclusions. "I'm going to write down those actions bluntly, very to the point of, 'You need to do X, Y and Z,'" she said.

2. Tell the story first, with data for support

Mico Yuk, co-founder of the Data Storytelling Academy and host of the Analytics on Fire podcast, said she borrows from Hollywood techniques in her approach to data storytelling. Yuk's BI Brainz website features this quote from her: "Start with the story and users will trust you. Start with technology and they'll forget you."

A report or a BI dashboard "may appear to be where a data story begins," she said in an interview. "But way before you get [to the data], what I'm feeding into your brain is actually where the story starts."

Yuk described data storytelling as "the art of taking data and adding what I call the touch, the feel, the smell and basically the emotion to it." Doing so brings data to life and "activates the senses," she added.

Her coaching framework on data storytelling has three main sections: what to ask about the data, what to write and what to visualize. By far, most of her clients want her to spend 80% of the training time on what to ask and what to write, Yuk said. Data visualization seems less important to them as a whole.

In data stories, Yuk encourages the use of "visual words" to bring images to mind and "visual action words" to suggest actions. These trigger emotional reactions more effectively than many of the everyday words used in business, she said.

Yuk invites students to feel the difference for themselves in coaching sessions. In one exercise, she has them draw the first images that come to mind when they hear common business words and then more emotionally laden ones. Words such as increase and decrease tend to prompt inanimate images, Yuk said. But, for example, when increase is replaced with boost, students draw "all kinds of fun, animate stuff."

3. Reduce data's overhead in data stories

Yuk and Lee Feinberg, president of consulting firm DecisionViz, seem to agree on at least one thing: where data belongs in a data story. In fact, Feinberg would rather not say data at all in this context. "Very rarely is it all about data," he said. "It's just storytelling. Or it's literacy. Or visualization. I think that by using the word data, it sometimes pushes people away."

Feinberg has been consulting and teaching on storytelling, analytics and visualization topics for more than 12 years; today, in addition to his work at DecisionViz, he's an instructional assistant in the applied data science master's program at the University of Chicago. He said the real challenge for data storytellers is how to give better presentations.

"You're trying to persuade people about a recommendation you have, or you're trying to educate people on an insight you have, and maybe you're just discussing what do [they] do next," Feinberg said. When he listens to a presentation, he doesn't care about the data itself. "What I care about is what it means."

But data is still involved, and it provides an important foundation for business decisions. The problem Feinberg wants to fix is the overhead involved in showing it as part of data stories. He sees that as an issue for data storytellers and their audiences alike: Creating custom data visualizations and figuring out how to read them are both too time-consuming, he said.

Feinberg claims to have developed a way to do those things faster and at a lower cost. Enough with the creativity and away with the made-from-scratch charts for every project, he said. His Design To Act methodology makes data presentations fit a template and adhere to a set of conventions for displaying various types of data.

This approach is intended to relieve dashboard developers and data storytellers from design questions and make them think about why they're creating a dashboard or a visualization in the first place. What's the message? What's the business problem to be solved? Feinberg's methodology also aims to help business stakeholders more readily see what the data means to them.

4. 'World building' through compelling narratives

Paulina Davila is vice president of workforce analytics and head of data visualization and storytelling at financial services firm JPMorgan Chase. She's enthused about the possibilities of data storytelling but said it's a skill that requires practice. "It really is a matter of practicing it to be more comfortable and for it to come naturally to you." Awareness that it's an important and valid activity is also required, she added.

JPMorgan Chase recruited Davila in 2022 for her data expertise and the marketing and advertising skills she acquired in previous jobs. Perhaps most relevant for data storytelling, she had a knack for creating marketable messages.

So long ago that she can't remember the source now, Davila was advised to closely observe engaging TV shows, movies or novels and dissect their narrative arcs. After doing so for a short time, she recognized how storylines developed, others were folded into them and characters emerged. Soon, she could apply those narrative techniques in data stories.

The concept of "world building" can also be part of data storytelling, Davila said. "You create a world -- you say what is true and what's not true, what exists or doesn't exist in this world. When you build companies, you create that [same] sense of a world. That's how you get people to get excited to work toward a goal."

In addition, she noted how a data story can be repeated by members of the original audience to others in their organization. Think of this as a "Starbucks story," one that's retold casually and socially -- in a conversation over coffee in the office, for example. Davila said it's a test of the original rendition's clarity and stickiness: Do the story's points endure, and do they still make emotional sense? Well-told stories are repeated effectively.

Davila can't give data storytelling examples from JPMorgan Chase because of confidentiality restrictions. But she did describe a barebones outline from her past role at an ad agency that later became a baseline for her storytelling approach. The outline listed three steps for a review of an ad campaign's performance by her team of BI analysts. Even those steps formed a basic story structure. First came a typical story's exposition, introducing the situation at hand: the team's review of the target KPIs. Then came the complication: The review showed the campaign had fallen short. Finally, the resolution: a recommendation based on the review.

5. Use an 'insight funnel' to frame and create data stories

Brent Dykes was working at a popular vendor of BI and data visualization tools when he had an epiphany: "Wait -- dashboards don't tell stories." A senior manager told him he didn't know what he was talking about. Dykes eventually left the company, which he asked not to be named. Since 2021, he has been founder and chief data storyteller at training firm AnalyticsHero. He's also the author of Effective Data Storytelling.

"I skew more toward the humanity of data storytelling," Dykes said. He added that, as human beings, we still have a central role in communicating our findings and insights about data. In telling data stories, "technology augments," he said. "It doesn't automate."

But data analytics "has really had a reporting mindset for a long time," Dykes said. In his view, a mindset shift is required to move from basic BI reporting to data storytelling.

The first step up is what he calls narrative reporting. It's reporting and interpretation, or "what-plus." But a narrative report is still structured as a report, not a compelling story, Dykes said. It might contain multiple story fragments, but there's no single story cutting through everything. The latter is what he sees as genuine data storytelling. "I'm not telling multiple stories at the same time," he said. "It's one story, one destination."

To get there, Dykes promotes the concept of an "insight funnel" to refine raw data into data stories. First comes data framing and exploration. The data that goes in the top of the funnel is examined through the lens of KPIs, strategic objectives and business goals. Some of it isn't usable or important, but what passes through that lens goes into reports or dashboards.

The next level down produces story framing. "We start to spot things. We start to see insights," Dykes said. At the lowest level of the funnel, data stories are clarified and created. Here, the data is explained, with its meaning made clear and recommendations given based on the data. "It's not just, 'Oh, here's an insight.' It's much more about persuasion, [and] taking action."

But insights are a crucial element of effective data stories, according to Dykes -- and a true insight isn't just an interesting point. "It shifts our understanding in an unexpected way," he said. "A boring story doesn't have that shift -- there's no drama. Then that [insight] causes us to say, 'We need to change our marketing.' Or, 'We need to change our product design.' There's a consequence."

6. Move toward a more interactive 'data storytelling 2.0'

While Dykes envisions a clear path from data to story, fellow storytelling pioneer Zach Gemignani takes a more philosophical approach. He shares Dykes' conviction about data stories focusing attention on meaningful data. But he questions some fundamental assumptions about visualizing data and thinking there's only one story to tell about a data set.

Gemignani is co-founder and CEO of analytics consulting, development services and software provider Juice Analytics. When he entered the consulting business in 2005, data visualization had just dawned. Data stories were one-way presentations; they provided a capstone to data analytics projects and leaned on the view that visuals carried every story. Gemignani calls that mode data storytelling 1.0.

"People just breezed in and assumed that good data storytelling came directly from visuals and data," he said. In this conception, the storyteller strides in with data analysis done and conclusions decided. But it leaves out one crucial element: the choices that data analysts and storytellers make along the way.

The first choice is selecting the data to use. Other choices are made in preparing the data, analyzing it and deciding on a data story. But does the data storyteller presume to have found the "correct" story? What happens if different data is selected or the data is prepared differently? Or if the same data set spawns different stories among an analytics team or the intended audience, perhaps with multiple storytellers?

A multistoryteller mode emerged as Gemignani's own thinking about data storytelling evolved. By the time his team at Juice began work on a self-service version of the company's Juicebox data storytelling tool, they had discarded the one-story assumption. The tool, initially released in 2020, was designed for "story finding." It lets users examine data, try out theories, sort through the possibilities and arrive at a story -- or multiple stories. Gemignani calls this data storytelling 2.0.

Not everyone shared his enthusiasm for interactive data exploration and story finding. When Nancy Duarte, a well-known data storytelling writer, speaker and consultant, viewed Juicebox, "she was befuddled," Gemignani recalled. "She said, 'I don't understand how you can tell a story if every time you click, the story changes or the information changes.'"

Juice has since added a conventional storytelling option to Juicebox for users who prefer less interaction and more focused, linear presentations. For some people, mixing in any amount of exploration instead of pure explanation destroys the story and its message, Gemignani said. In such cases, a "curator," as he put it, examines the data and decides on a story. However, the interactive exploration option remains for the storytelling 2.0 users he originally envisioned.

7. Curation choices drive data storytelling

Like Gemignani, Andy Cotgreave contemplates issues about the data storyteller's choices. Cotgreave might be best known among attendees of Tableau user conferences as the glittery-suited emcee for the Iron Viz event, a data visualization competition. But his regular job is senior data evangelist at Tableau, which is now owned by Salesforce following a 2019 acquisition.

Cotgreave said a good data story begins to take shape with the most basic choice: which data to use. That can reach all the way down to the choice of the raw data points collected for analysis. "Storytelling is decided to an extent when you store the data," he said. What data is stored and what's left out affects the final narrative.

But the data storyteller then has the obligation to choose and curate the stories, Cotgreave said. "You could choose to show the average sales of A and B from this data set that was collected, and you can tell two very different stories depending on which average you choose." For example, a mean average might show one product outselling another, but the order might be reversed if the median average is used instead.

"There is no 'single version of the truth,'" Cotgreave said. "That's a selling tool [for BI vendors]." As he sees it, data collection is a curated set of decisions about what data you use. Analyzing it and deriving insights to include in data stories is a curated interpretation of the facts contained in the chosen data set.

To create a useful data story, Cotgreave said, it's best to first agree on the most fundamental decisions: What questions will be asked? Which data will be used? What's the story? In his view, the data storyteller ultimately decides.

Ted Cuzzillo is a freelance technology journalist who focuses on data storytelling and analytics and has written a blog called Datadoodle since 2007.