DiscoverDataScience.org

FIND A PROGRAM
1
2
3
4

Data Storytelling: How to Master the Core Skillset of Data Science

Data storytelling is sometimes referred to as “the last mile in data analytics” because the narrative used to engage an audience, if told effectively, can transform them from passive listeners to active participants. 

The skill set needed to tell a story with data includes a variety of soft skills, such as creativity, communication, and the ability to craft a narrative that will be meaningful to your audience. 

These non-technical skills ensure that all the hard work and technical knowledge that goes into extracting insights from a data set actually results in action. 

In a DataFramed interview — DataCamp‘s official podcast — the senior director of data strategy at Domo and author of “Effective Data Storytelling,” Brent Dykes, emphasized the perils of neglecting to tell the story behind the data. 

“We’ve done a lot of work to find an insight; all of the capturing of the data, processing of the data, organizing, analyzing, we build a model, and then we come away with this insight. All of that work will be for naught if we’re unable to communicate it clearly to other people and have them consider it and act on it,” he told DataFramed host Adel Nehme. 

FIND SCHOOLS
Sponsored Content

What Is a Data Storyteller? 

A data storyteller is a data professional who can craft a compelling narrative around their data to move an audience to action. Depending on the data set and the objective of the project, the audience may consist of decision-makers and stakeholders in an organization, consumers, or the public at large. 

Data scientist Cole Nussbaumer Knaflic’s book “Storytelling with Data” was written for anyone attempting to communicate something to an audience using data. This includes “analysts sharing the results of their work, students visualizing thesis data, managers needing to communicate in a data-driven way, philanthropists proving their impact, and leaders informing their board.” 

In other words, data storytellers present to virtually any group of people looking to find answers to questions or solutions to problems. From Fortune 500 companies to public health agencies and nonprofits, data storytelling is used in a variety of industries. It can communicate sports analyses, findings related to health informatics and the health-care industry, areas for improvement in government policy, risks to public health, gaps in educational services, opportunities for sales and marketing teams, and a host of other solutions. 

For example, Ben Wellington, a quantitative analyst at Two Sigma and the founder of I Quant NY, conducted a project using data from NYC Open Data to determine where people get out of taxis in New York City. His research showed that the behavior of taxi riders depended on the direction in which they were going. He presented the data to his TEDx Talk audience by connecting it to a shared experience. 

“If you’re coming in on 46th — you’re going west to east — you seem to get off on the west side of the avenue. Why? You ever get a cab stuck in traffic? You just get out? You can actually see it in the data. […] And this is interesting, right? If you’re trying to do advertising in the district, you might want to know where to welcome people, where are people getting out most often. We can start to study this with our public data.” 

Talented data storytellers go beyond visualizations and dashboards in their appeals to their audience. They develop presentations that revolve around key insights, curated specifically to emphasize the most salient and urgent points, and deliver them in a way that resonates with their target audience. 

Why Is Data Storytelling Important? 

Data science professionals analyze and interpret data for a living, but for everyone else, raw data without context is similar to a story without a premise or structure. It is generally hard to follow, of little value, and is likely to generate more questions than answers. Creating a data story allows data scientists to provide context and bridge the gap between what the data tells them and what the audience needs to understand. 

A data story is also able to make the data relatable and relevant to the audience, relying on a fundamental aspect of human nature: emotion. When the audience feels connected to the data on an emotional level, they care about the message and feel inspired to act. 

Now that the audience understands the data and wants to take action, the data storyteller is able to set expectations for the proposed solution and outline next steps by creating a clear call to action, which we’ll explore in greater detail later in the article. 

Ultimately, data storytelling has the power to influence leaders, galvanize teams, transform organizations, bolster innovation, and overcome resistance to change because the story around the data is more effective than the data alone. 

The story ensures that the audience comprehends the key takeaways, connects emotionally, and is driven to act expediently on its clear and direct call to action. 

How to Tell a Story with Data 

In his interview with DataFramed, Dykes warned that the term “data storytelling” has been misused and misunderstood, noting three common misconceptions about data storytelling. 

  1. Data storytelling is a synonym for visualizations. 
  2. Dashboards tell data stories. 
  3. Text (in the form of annotations and commentary) on visualizations turn them into a story. 

While visualizations, dashboards, and annotations have their places in a strong data presentation, these elements are not data stories in and of themselves. They are components of a much larger narrative. 

Below are the common steps to building am evocative data story. 

Step 1: Get to Know Your Audience 

While there’s no single narrative structure for a compelling data story, data scientists agree that the best way to tell a story with data and analytics is to first understand your audience. What motivates them? What problems do they want to solve? Which of your insights will pique their interests and resonate most with them? 

From there, an effectives data storyteller will use this problem — or, in narrative terms, this conflict — to hook their audience. Just like any story, whether told for entertainment, context, connection, or retention, a data story must capture the attention of the audience. 

As data analyst and founder of Trending Analytics Hana Khan told Discover Data Science, an important part of data storytelling is knowing your audience. This, she said, makes the job much easier because when you know what your audience is concerned about, or what motivates them, you can identify which story from the data you should be telling. 

Step 2: Determine Your Key Message 

Whether the objective of your data project is to increase sales, affect policy, or implement organizational change, your data story needs a key message — a clear takeaway upon which your audience can act. 

In her TEDx Talk, titled “Why storytelling is more trustworthy than presenting data,” CEO and Chief Storyteller of Eber Leadership Group Karen Eber says a good data story builds an idea and “helps [the audience] see something they can’t unsee.” 

This idea is the key message and the focus of the story. 

In her example, Eber recounted the tale of a college advisor who noticed that a high percentage of the college’s students with autism weren’t graduating. When the advisor presented her findings to university officials, she centered her presentation around a student with autism named Michelle. 

She spun a detailed and emotional narrative about Michelle’s early challenges, the adjustments Michelle and her advisor had made to her learning experience, and her resulting success before sharing the data with her audience. The advisor ensured that the statistics she presented and the tone of her story supported her key message, which was: The university needed to make changes in the way they served their students with autism. 

Good data storytellers know their key message and support that core message throughout the narrative. 

Step 3: Select Your Data to Address the Needs of Your Audience 

It’s important to present only the most relevant data to your audience. 

According to Dykes, “Any mental effort the audience members waste on extraneous items reduces their capacity to focus on your core message.” 

The ability to extract these key findings is a valuable data science skill. As Khan notes, this is not simply cherry-picking insights to align with your audience’s hypotheses, but instead, selecting those that will resonate with your audience. 

“Other skills that are helpful include being able to think analytically and critically so you’re able to identify the relevant insights from the data that will help you tell that story, rather than just talking about the data and overwhelming your audience with information that may not be relevant to them,” Khan said. 

Step 4: Develop Your Data Visualizations 

The insights a data storyteller chooses to share will vary depending on whether they are addressing other data scientist or a nontechnical audience, but once they’ve made this determination, and consequently have decided on the narrative framework, they can create visualizations that support each of these insights. 

According to Ashley Fell, social researcher, TEDx speaker and head of communications at McCrindle, our brains process visuals 60,000 times faster than text. This statistic alone makes it clear that effective visuals have a significant impact on a data story. 

But not just any visualization will have the desired impact. Good data storytellers are intentional with their visualizations. 

Knaflic covers six key lessons for data visualization in her book: 

  1. Understand the context. 
  2. Choose an appropriate visual display. 
  3. Eliminate clutter. 
  4. Focus attention where you want it. 
  5. Think like a designer. 
  6. Tell a Story. 

In the final lesson — Tell a Story — Knaflic uses an example of product pricing recommendations to show how visualizations can guide an audience through a data story. 

“By drawing our audience’s attention to the specific part of the story we want to focus on — either by only showing the relevant points or by pushing other things to the background and emphasizing only the relevant pieces and pairing this with a thoughtful narrative — we’ve led our audience through the story.” 

Both Knaflic and Khan acknowledge that although there are a variety of graphing applications and presentation tools that will help data scientists and analysts tell their stories effectively, these tools are supplementary to the crafting and delivering of a compelling narrative with a clear call to action. 

While knowledge of Excel, Python, or other data visualization tools are critical for preparing your charts and graphs, these tools can’t interpret the meaning behind the data or provide you with the structure for your story. 

Step 5: Structure Your Presentation 

The structure of your data story will depend on a number of factors, including whether you’re delivering your story in-person or via a written format, the diversity of your audience, and the time you are allotted to present, among others. 

Dykes suggests that flexible and adaptive data storytellers have the advantage, as each presentation scenario will be different. When they must deviate from data storytelling standards, he advises his readers to “do so in an informed, calculated manner, knowing what you’re sacrificing to achieve a specific aim with your storytelling.” 

We discuss data story structure in more detail below. 

Step 6: Tell Your Story 

Each of the data science experts we’ve mentioned agree that storytelling is exponentially more powerful than presenting raw data to an audience, and many of them point to the neuroscience behind storytelling. 

Eber and Dykes, for instance, refer to neural coupling, which is a phenomenon in which the brain activities of a storyteller and a listener synchronize. 

Neural coupling, Dykes explains in the chapter on the psychology of data storytelling, was discovered by neuroscientist Uri Hasson, who “found the stronger the coupling between the two parties, the better the communication and the deeper the audience’s understanding.” 

As Eber explains, data doesn’t change people’s behavior; emotions do. Through their research on decision-making, neuroscientists have discovered that humans begin to make decisions at a subconscious level as the amygdala processes emotions. It isn’t until after this subconscious decision-making occurs that we begin to apply logic to the process. 

This is essential information for data storytellers, who should understand that the emotional core of their stories have a much stronger impact than one might imagine. 

The Structure of a Data Story 

“We’re all familiar with stories, and we love stories. And a very common structure for stories is that they have a beginning, middle, and end,” Khan said. 

It’s the job of the data storyteller to figure out what goes at the beginning, middle, and end of the story. 

Khan suggests that data stories tend to have a common framework based on the fact that data analysts come across problems all the time. These problems present the conflict, or the hook, that makes the audience keep listening to find out how this conflict will be resolved. 

She offers an example of data storytelling for business, in which she advises data storytellers to put the problem in the hook of their story, at the beginning. 

“If you’re presenting to business stakeholders, I recommend you talk about a problem that is directly related to the business so that the audience thinks, ‘Oh, that is something I’m worried about, and now you’re saying there’s a problem.’” 

But, she says, don’t give them the solution right away. Just like any other story, the data story should contain an element of suspense. 

The middle of the story sets up the context. At this point, you provide the background information and dive into the insights and the journey the data team made to extract them and come up with a solution to the problem. 

This is also where Khan recommends you reveal the data-driven solution and make your recommendation. 

Once you’ve reached the end of the story, you’re ready to tie it all together. Summarize the main idea, and circle back to the problem. Reiterate your recommendation and the call to action and remember to include a timeline or a sense of urgency — if there is one. 

Don’t worry about repeating yourself. 

According to Khan, “Repetition helps get a point across. This is the most important part of your presentation — the call to action — so I don’t mind repeating it.” (In addition to a comprehensive Data Presentation Course, Khan offers a Free Data Presentation Roadmap for download on her website with more tips and best practices for making “effective and confident data presentations.”)

Dykes offers three narrative models for data storytelling. 

The first — and simplest — is Aristotle’s Tragedy Structure, which consists of a complication and an unraveling. 

The second narrative model is Freytag’s Pyramid, which includes additional elements to Aristotle’s structure, most notably the inciting incident and the resolution. 

And the last and most complex of the three models is Joseph Campbell’s Hero’s Journey, which Dykes deemed “too complicated for assembling data stories.” 

From these three models, Dykes developed and proposes a four-stage narrative structure that comprises the following: 

  • Setting: This stage presents the background and hook. 
  • Rising Insights: This stage provides the audience with supporting details. 
  • Aha Moment: The Aha Moment is the point at which you present your key insight. 
  • Solution and Next Steps: At this stage you present your solution and a clear call to action for your audience. 

For a more intuitive narrative flow, Knaflic suggests the data scientist consider certain factors when deciding how they want their data story to unfold. Details such as the knowledge level of the audience, the level of credibility the storyteller has with their audience, the call to action, and the level of input needed from the audience will all affect the narrative structure of a data story. 

The Importance of Effective Data Visualization 

In their book, “The Big Book of Dashboards: Visualizing Your Data Using Real-World Business Scenarios,” authors Steve Wexler, Jeffrey Shaffer, and Andy Cotgreave aptly state, “You can’t, in a single chart, answer every possible question or comparison. What you do need to do is assess whether the chart you do choose answers the question being asked.” 

It’s not enough to simply add a chart to your presentation or report. It must be designed to convey the insight in an effective way. 

Data storytellers should be considering their audience from the beginning — including before they create the visualizations to accompany their story. Titles and labels are important in conveying the message and ensuring that the audience understands what the chart or graph is relaying about the data and that they care about the information. 

While visualizations are not the whole story, they go hand-in-hand with narrative and add clarity to your story. They prevent the audience from losing focus during a presentation by guiding them to the insights that correlate to the part of the story you’re telling. 

Poorly conceived or executed visuals can kill your data story. 

Khan explains the advantage of using a powerful data visual rather than a simple table in a presentation. 

“[The audience] is able to understand the insights better from a data visual versus a table because their eyes are trying to evaluate every number [in a table], and you’re making it harder for them to see the insight. 

When you add this friction, you could potentially lose their attention because if they see a slide with a huge table, they may just skip it altogether. Or they may actually shift their focus to the numbers because in their heads, they’re trying to do the comparison.” 

Dominic Bohan of StoryIQ reiterates these viewpoints in his 2019 TEDx Talk “Turning Bad Charts Into Compelling Data Stories.” 

Bohan advocates for ruthless minimalism, human-friendly charts, and clear takeaways. 

According to the Australia-based data storytelling trainer, “Data is useless unless human beings can interpret, analyze and understand it, and use it to drive action.” 

Bad charts make it difficult, if not impossible, for people to understand the data they are being shown. 

Search Engine Watch lists effective storytelling visualizations, including: 

  • Bar Graphs 
  • Bubble Charts 
  • Histograms 
  • Infographics 
  • Line Charts 
  • Maps 
  • Pie Charts 
  • Scatter Plots 

These are just a few of the visualizations you could use to support your data story. Just remember that certain types of charts are more effective than others for specific types of data and communication. 

FIND SCHOOLS
Sponsored Content

Data Storytelling Courses & Training 

Traditional data science courses teach many of the soft skills you need to become a great data storyteller. 

There are also free resources available on the internet, including those that can help data scientists with data visualizations. 

Interestingly, some experts even suggest that folks with liberal arts backgrounds have a skillset uniquely suitable for bridging the gap between analytics and action. 

“In an IT department, you are most likely to find these [strong communicators who possess both business acumen and IT savvy] in the positions of business analysts, and many of them (as in the past) will be capable communicators who hail from liberal arts backgrounds,” writes Tech Republic contributor Mary Shacklett. 

Data Visualization Courses & Certificates 

Universities across the United States and around the globe offer courses and training for data storytellers. You can find individual courses, certificate programs, and broad data science bachelor’s and master’s degree programs with data visualization concentrations to suit your interests and career goals. 

Johns Hopkins University 

The Data Visualization course, which is taught primarily as part of the computer science program at Johns Hopkins University, covers the theory and practice of creating visual representations of large data sets. 

  • Course Number: 605.662 
  • Delivery Mode: Online 
  • Prerequisites: Experience with data collection and analysis in data-intensive fields or background in computer graphics 

University of Illinois 

Offered on the Urbana-Champaign campus, the University of Illinois’ Data Visualization course covers appropriate and effective chart selection and construction, dashboard organization, and user-focused design. 

  • Course Number: CS 416 DVV 
  • Delivery Mode: Online 
  • Prerequisites: CS 225 

University of Illinois School of Information Sciences 

The IS 445 Data Visualization course at the University of Illinois Urbana-Champaign gives students a foundation in constructing communicative visualizations, modern software for data visualization, and techniques for collecting and interpreting data through effective visualizations. 

  • Course Number: IS 445 
  • Delivery Mode: Online 
  • Prerequisites: NA 

MacQuarie University 

The Department of Actuarial Studies and Business Analytics at MacQuarie University in Australia offers a Data and Visualization for Business course that provides students with comprehensive business case studies. Using Linux, SQL, and a variety of other data analysis tools, the course gives students a foundation in data visualization. 

  • Course Number: BUSA8090 
  • Delivery Mode: On campus (held weekly) 
  • Prerequisites: Admission to MActPrac or MAppStat or MBusAnalytics 

University of Washington Professional and Continuing Education 

The Certificate in Data Visualization from UW Professional and Continuing Education is an 8-month certificate program that consists of three courses. The tuition fee for the program is $4,077 and students can take certain courses without enrolling in the full program. The program explores data visualization and data storytelling for a variety of audiences. 

  • Courses 
  • Data Visualization Theory: A Practical Introduction 
  • Data Visualization Presentation: Dashboards & Storytelling 
  • Decision-Making Through Data Visualization 
  • Delivery Mode: Online with synchronous meetings 
  • Prerequisites: NA 

eCornell 

eCornell, Cornell University’s external education unit, has designed a Data Visualization in Tableau certificate program tailored to a variety of professionals who work with data, including marketers, project managers, business and data analysts, managers, consultants, and developers — among others. 

In this 2-month program, students learn to create and improve data visualizations that are tailored to a specific audience and crafts a powerful data story that drives action. 

  • Courses 
  • Creating Data Visualizations with Tableau 
  • Enhancing Data Visualizations with Tableau 
  • Telling a Data-Driven Story with Tableau 
  • Elective Course 
  • Delivery Mode: Online 
  • Prerequisites: NA 

Harvard University 

Harvard’s Data Science: Visualization course is an introductory-level course that belongs to the Professional Certificate in Data Science series. It’s an 8-week course that covers basic data visualization principles and exploratory data analysis using R’s data visualization package, ggplot2, and three real-world examples. 

  • Course Number: NA 
  • Delivery Mode: Online 
  • Prerequisites: An up-to-date browser is recommended to enable programming directly in a browser-based interface. 

Featured Master’s Programs 

For those interested in completing a graduate degree in data science, we offer a comprehensive list of online master’s in data science degree programs. Find the one that’s right for you. 

Return to Discover Data Science Articles