The Future Of Data Storytelling Is Augmented, Not Automated

As evidenced by the Palawa oral traditions of the Tasmanian Aboriginals and the Chauvet cave drawings in Southern France, storytelling has been an integral part of our humanity for tens of thousands of years. Wired for Story author Lisa Kron stated, “Story, as it turns out, was crucial to our evolution—more so than opposable thumbs. Opposable thumbs let us hang on; story told us what to hang on to.”

And yet, in business settings, it had been mainly relegated to the water cooler or lunchroom—not the boardroom. However, its transformative potential caught the attention of insightful leaders like Apple founder Steve Jobs. In 1994, he is quoted as saying, “The most powerful person in the world is the storyteller. The storyteller sets the vision, values, and agenda of an entire generation that is to come.”

Today, organizations are increasingly recognizing the importance of storytelling in business. Whether stories are used to communicate a strategic vision, core values or the results of a new initiative, they help leaders to connect with others on an emotional level. It’s hard to ignore its potency as a communication tool. Telling a story can transport an audience and make your information more engaging, memorable and persuasive.

The influx of data and artificial intelligence

Two significant shifts are impacting storytelling in business. First, as organizations amass greater amounts of data, it’s becoming an increasingly common component in most communications. However, making sense of the numbers, finding actionable insights and communicating them effectively can be difficult. As companies stockpile more and more data, their data investments won’t generate adequate returns if insights aren’t being used to enhance decision making. Data storytelling helps companies unlock this potential by skillfully combining data, narrative and visuals into compelling data stories that inform decisions and drive positive change.

Second, the rapid advancements in artificial intelligence are beginning to transform how businesses operate. Whereas human beings struggle to process vast quantities of data quickly and accurately, machines can handle it with ease. With the volume and complexity of data increasing each year, AI will play a vital role in helping companies turn data into action. When building data stories takes time and effort, some people may wonder whether AI tools can be used to automate data storytelling.

As someone who has authored a book on the topic and focused on it for over ten years, I see four main reasons why data storytelling can’t or shouldn’t be fully automated.

1. Oversimplification of what data storytelling is

A prevailing misconception about data storytelling is that it is simply about visualizing data in charts and graphs. As a result, the chart choice and visual design aspects receive most of the focus. The narrative element becomes marginalized and reduced to having good headlines or annotations. Meaningful data stories require a solid narrative structure that weaves multiple data points into a cohesive story. Various narrative techniques, such as humanizing the data, will also amplify the emotional impact of your data stories. When you neglect developing a strong narrative, you may create appealing data visualizations but weak data stories.

What many people call data storytelling, I call narrative reporting. Data storytelling and reporting are different and don’t serve the same purposes. While they may look similar to data stories, narrative reports follow the same descriptive reporting format as traditional reports but with some added context and interpretation. On closer examination, you’ll discover that a data story offers richer information because it is based on a deeper diagnostic analysis (why) rather than just a descriptive summary (what).

More importantly, data stories present their information in a narrative structure rather than a reporting structure. Narrative reports only provide incomplete story fragments, whereas data stories offer complete narratives. To use a fast food analogy, narrative reporting offers disjointed menu items, whereas data storytelling provides a complete combo meal.

If the goal of data storytelling were to generate informative charts with added commentary based exclusively on what’s visible in the data, AI could be used to generate these basic narrative reports. Today, many analytics platforms offer “data storytelling” capabilities that leverage Natural Language Generation (NLG) to add descriptive text to charts. However, this automated commentary lacks the crucial contextual understanding that falls outside its known datasets. Just like an analyst will only be able to provide meaningful interpretations if she understands her business’s priorities and nuances, the same applies to AI. While it can offer some value with narrative reporting, it will be more challenging for AI to craft comprehensive, compelling storylines that are more than just information summaries.

2. Storytelling is a uniquely human skill

Human beings can’t match the processing power or speed of large learning models (LLMs) in working with large data sets. We are prone to fatigue and error, especially with repetitive data tasks. Many tasks in the analytics process, such as data cleansing, preparation and analysis, can be enhanced by machines. However, what humans can offer AI is an innate strength in storytelling. People can outperform AI in the following storytelling-related areas:

• Creativity and imagination: Humans can creatively explore the data and communicate key findings in inventive ways that engage audiences.
• Complex problem-solving: Humans are skilled at solving ambiguous problems and developing novel solutions.
• Emotional intelligence: Humans understand other people’s feelings (empathy) and how to convey emotions, which are integral to effective storytelling.
• Context awareness: Humans can look beyond the data to gain more context into internal and external factors influencing the numbers.
• Judgment: Humans can navigate nuanced business scenarios and determine when something doesn’t feel right or ethical.
• Adaptation to audiences: Humans can tailor stories to the needs and interests of specific audiences. They can also adjust in real-time as audiences react to their stories in various ways.

Without these essential human qualities, AI’s ability to create and tell meaningful data stories will be limited. While AI excels at recognizing anomalies, patterns and trends in complex data, it still lacks the human judgment to know what should be prioritized and how best to package the content for specific audiences.

Automated data stories will face a similar challenge to what automated alerts have experienced. Whenever business users get a high number of irrelevant alerts, they perceive them as nuisances rather than useful notifications. Business users will either disable or ignore them. Automated data stories will run into the same issue if they aren’t relevant and meaningful.

3. Transparency and trust are essential to adoption

In the fourth century BC, Greek philosopher Aristotle identified three main persuasive appeals in his Rhetorical Triangle: logos, pathos, and ethos. While data storytelling blends logos (facts/reason) with pathos (emotion/narrative), the importance of ethos (character/integrity) can’t be overlooked. The credibility or authority of the data storyteller and their sources are essential to data stories being not just heard but trusted and adopted.

One of the main reasons why we share data stories is to convince an audience, using data and narrative, that a change is necessary. This often means we’re challenging the status quo and likely to encounter audiences resistant to the new information and a proposed new direction. When their viewpoints and assumptions are being challenged, audiences will be skeptical of what’s being shared with them. This is where ethos, through transparency and trust, builds a bridge between the storyteller and the audience to overcome this resistance. Unfortunately, AI’s mysterious ‘black box’ approach to finding and sharing insights makes it difficult for audiences to trust and embrace automated data stories.

In addition, one of the greatest threats to AI’s trustworthiness is the possibility of hallucinations where an AI model invents false information and presents it as fact. In data storytelling contexts, where the AI model may not be appropriately tuned for the analysis—either underfitted or overfitted—along with the presence of flawed data quality and insufficient context, there’s more potential for AI to produce these erroneous interpretations that could erode people’s trust in its automated data stories.

4. Not everything that can be automated should be

Even though it may be possible to automate something, it may not always be the most prudent course of action. Let’s imagine your fiancé decided they wanted to fully automate your wedding. Across the different planning, coordination and execution aspects of the wedding, you could employ AI and technology to streamline and enhance the entire experience.

• Planning: AI could build a plan for all aspects of your wedding, make necessary bookings, establish a gift registry and keep expenses within a designated budget.
• Invitations and guest management: AI could create and send out digital invitations, manage the guest list and prepare seating arrangements.
• Ceremony and reception: AI could produce the wedding vows, create a virtual reality reception with fashionable decorations and attire, officiate the ceremony and DJ the music. Based on submitted photos, AI could turn them into glamorous, professional-looking photos and videos of the entire wedding party and participants.
• Post-wedding: AI could send personalized thank you emails to all guests and assemble photo albums and highlight reels of the virtual wedding and honeymoon.

Doesn’t this sound amazing—or does it? While aspects of this automated wedding idea may sound appealing to some people, it would rob them and their loved ones of a momentous life experience. The special connection that comes from the joys and pains of orchestrating and participating in such an event would be lost and replaced with fleeting artificial memories.

Similarly, letting AI automate all aspects of data storytelling would mean you would lose opportunities to deepen your understanding of the business and build stronger relationships with key audiences. Aside from whether automated data stories are feasible, data storytelling is often the most interesting and rewarding part of the entire analytics process where insights are finally translated into action. Relinquishing this responsibility entirely to machines could lead to more long-term pain that outweighs any short-term gain.

The path forward is augmented data storytelling

The discipline of data storytelling is truly a blend of art and science. On the art side, you combine elements such as business acumen, audience empathy, narrative formation, visual design and messaging. On the science side, you benefit from understanding analysis, psychology and behavioral economics. This unique mix of skills and knowledge makes data storytelling hard to master and even more challenging to automate.

By its nature, AI will excel more with the science side of data storytelling, which represents more technical and structured tasks. On the other hand, humans have distinct advantages on the art side, which reflects the more creative, nuanced and empathetic aspects of crafting data stories. By having AI augment humans (or humans augment AI), we create a powerful partnership that increases the volume of data stories and significantly elevates their quality beyond what either could produce alone. Looking at the individual steps in the data storytelling process, you can see what each side brings to the combined equation.

As the table above shows, humans and AI have different strengths and weaknesses across the data storytelling process. While humans are accustomed to crafting and telling stories, doing so with data presents new challenges. With the emergence of AI, we now have a potential partner that can augment how we find and tell stories with data. Even though human beings have been developing their storytelling skills over thousands of years, the rapid pace of AI innovation means we may achieve more with less effort. Data storytelling is a responsibility we shouldn’t mind sharing but also one that humans should never surrender—it’s engrained in our DNA and essential to our success.