Data Storytelling Is the Conduit for Modern Data Literacy, Nightingale
All the images in this article are taken from a keynote about humanizing AI that I delivered in June 2022 at the London Tech Week. This article has been written to expand on the point of view on humanizing AI that I presented in that keynote. Regardless of the publication date, this article was written during my time as an IBM employee, and it’s meant to celebrate my accomplishments in data and AI design and share my story with other leaders in the field.
No data storytelling, no understanding of AI, no innovation: it is that simple.
As part of a team of data scientists, I’ve watched many organizations fail to scale their AI pilots due to a lack of data storytelling skills. Surprised? Well, you shouldn’t be. Not using stories to explain and contextualize the outcome of AI solutions results in executives struggling to grasp their business value and not making any investments to advance them. I’ve experienced firsthand the consequences of failures in communicating data.
On many occasions, the charts and analysis my team presented to explain the functionality of ML models and algorithms left the users wondering about their utility and reliability. A wrong decision based on misunderstood data could result in revenue erosion and angry customers. To trust and adopt AI, users and business stakeholders must understand the implications of the data produced by the AI systems on both the organization and people.
Having spotted this gap repeatedly, I worked on infusing data storytelling throughout AI projects, from strategy to execution. I started by repurposing my role as a data journalist from merely designing data visualizations for AI tools to developing and communicating data and AI strategies.
Today, many organizations want to become fully AI-powered by implementing systems that improve customer experience and accelerate, optimize, and innovate business processes. However, the fast pace of technological advancement doesn’t go hand in hand with employees’ skills. Insufficient proficiency in reading, interpreting, synthesizing, and communicating data — competencies that definedata literacy — prevents organizations from capturing AI’s full potential.
Let’s be honest: deciphering the value of machine learning models is still a mystery to many business leaders. Trustworthy AI is a priority for many organizations, but often only data scientists can wrap their brains around stats describing models’ fairness or quality. Reading a confusion matrix is still, to most professionals, a challenge equal to codifying the Rosetta Stone. Not to mention translating models’ performance metrics into business KPIs. Forget about it!
There is still a chasm between business strategy and technology among organizations caused by a lack of data literacy. According to a data literacy survey by Accenture, only 21 percent of employees feel confident in their data skills.
It’s time to leave behind the assumption that data literacy is merely analyzing data and plotting charts. We need to combine different skills to produce stories that transform data into actionable knowledge and unfold the AI’s complexity. To progress with digital transformation, leaders must rethink the role of data literacy, which has become a fundamental requirement to fathom the world through the new gnoseological paradigms produced by our computational culture.
We are no longer dealing with only numeracy or graphicacy. Don’t get me wrong: of course, these skills set the foundations of data literacy. However, the automated systems that permeate our lives have exposed the workforce to much more complex challenges that call for new lenses to interpret data: which are no longer only a massive amount of structured and unstructured information, but also the fuel and outcome of the AI systems that people use or are affected by.
So, how could leaders solve this conundrum? There is no “one-size-fits-all” approach. Yet, by working on data and AI projects, I found that data storytelling functions as an enabler of data literacy.
During my time at IBM, I initiated many efforts to improve data literacy across the company’s business units and innovate people’s perception of data by bringing them closer to code, algorithms, and numbers through visual stories that read data but speak humans. Eventually, the initiatives I led culminated in the launch of the company’s first data storytelling-certified learning program.
Whether or not enterprises will succeed with digital transformation depends on their workforce’s ability to transform data into new opportunities, so leaders, you’d better act! Here are a few steps you might want to follow, and I’ll tell you why.
The Data Literacy Projectby Qlikq — a software company— reports that enterprises that have higher corporate data literacy scores can have $320-$534 million in higher enterprise value. Yet, few companies invest in data literacy efforts. According to Gartner’s 7th Chief Data Officer (CDO) Survey, only just over half of respondents (53 percent) said that they invested in the area of data literacy during 2021, while less than a third (29 percent) reported having successfully seen “ROI from data and analytics investments.”
Part of the problem lies in leaders making isolated investments in analytics teams instead of approaching data literacy from a holistic perspective: they often build data science communities and programs, but don’t implement a data literacy strategy addressing the entire workforce. Leaders should disperse data skills across business units by decentralizing the expertise of technical practitioners in data science, IT, analytics, or CDO teams.
Yet, making data competencies fluid also includes investing in education. Fostering a shift in how data is perceived and evolving the skills needed to become data literate leads people to discover new pockets of growth and generate new ideas to achieve the company’s strategic objectives more effectively through analytics.
Given that the vast majority of people — from HR to Finance, from Design to Sales — work, directly or indirectly, with data and algorithms, data literacy should be incorporated through learning programs suited to the challenges employees face in today’s ever-changing digital enterprises.
Training employees on unrelated technical skills like data analysis, coding, and data visualization doesn’t necessarily bring data and algorithms into their workflows: it’s like learning a language’s grammar without putting it into practice in everyday life by writing and speaking. By crafting data into narratives, people give purpose to the skills they learn and articulate them under a well-defined intent: communicating.
Data has become our tool for communication and expression. This is why we should perceive data as a language rather than merely a collection of digital information. To see clearly through the world and codify its dynamics, everyone should be proficient in the language of data.
As with language, where rhetorical thinking gives strategic purpose to the words, in the same way, rhetorical thinking applied to data gives strategic purpose to it. Thinking rhetorically about data brings us to discern the available means of analytics and combine them with ethos, logos, and pathos —data expertise, logical arguments, and emotions— to argue, communicate and persuade. MIT professor Catherine D’Ignazio and research scientist Rahul Bhargava include arguing with dataamong the abilities defining data literacy, recognizing as crucial being able to “use data to support a larger narrative that is intended to communicate some message or story to a particular audience.”
For over two years, I was in charge of curating and managing an internal newsletter to report my team’s achievements. I used that projectto convert the traditional newsletter format into a data-driven one. I experimented with creative ways to uncover overlooked stories hidden in data by leveraging Natural Language Processing to extract deeper meaning from the text of the stories I published. Making data less intimidating with aesthetics and rhetoric helped non-technical experts approach the realm of analytics, while revealing the human side of data to data scientists and engineers.
Data storytelling is a means leaders should employ to promote the rise of what Tsedal Neeley and Paul Leonardi define as a digital mindset: “a set of attitudes and behaviors that enable people and organizations to see how data, algorithms, and AI open up new possibilities […].” In fact, by intriguing people with unusual visual stories, the newsletter got more than 20K subscribers, raised cross-disciplinary collaborations, and spread new ideas on employing data and algorithms in mainstream business operations.
Crafting data stories that visually narrated the what, why, and how of AI models proved crucial to complete the hard work of data scientists and connected algorithms to the users’ daily experience through stories that they related to. However, visually explaining AI solutions is not enough.
To capture value from data and AI, leaders must develop a clear vision that aligns the technology’s intents with business and people’s needs. This is not an easy task, given the gap between AI and business: that’s why data storytelling is critical.
By researching this realm, I expanded the purpose of data storytelling and transformed it into a strategic tool to frame AI use cases through the lens of people’s needs. By combining data design with design thinking, I transformed workshops’ insights into data and data into visual stories allowing both technical and business stakeholders to gain a common language to translate complex analytical scenarios into a clear vision for their AI strategy. Elevating data storytelling to a strategic tool sparked a collaboration with the IBM Design Program Office that gave birth to a new design thinking framework that fundamentally changed how the organization addressed data and AI problems with a human-centered method.
By building bridges between unrelated domains, stories allow different experts to open up a dialogue on data to understand why and how to embed it into the fabric of their organization.
To bring analytics to the heart of the enterprise strategy, we must make an effort to weave data into people’s daily work through stories. As one of my favoriteHarvard Business Reviewarticles reads: “The ability to present data insights as a story will, more than anything else, help close the communication gap between algorithms and executives.”
We can’t improve how data circulates in companies without getting rid of guardrails that restrict its domain exclusively to analytics teams: this holds organizations back and can hamper innovation. Leaders should combine data scientists and engineers with professionals who, by crafting stories, can activate data into the business dynamics.
To innovate mainstream workflows and develop new approaches to analytics, we should try to infuse data storytelling into every project we work on, independently from the role we cover in the organization and whether or not we are data scientists.
Over the years, I collaborated with developers, data scientists, designers, and even executives, to expose me to their approach and expose them to mine. For example, I started developing interactive narratives adopting techniques like scrollytelling to make quarterly reports more engaging, instead of using traditional formats like reports, slides, and spreadsheet tables. I positioned storytelling at the core of the development of a new series of sales and revenue dashboards to let stories drive the understanding and design of the data the executives wanted to see to efficiently run their business. I even brought data storytelling into the software design team, where we explored how it could reinvent UX practices.
With time, the adoption of data storytelling raised people’s interest and curiosity. Many started asking for mentorship and coaching sessions. I understood it was time to develop an official learning program.
Although practicing is vital, a well-structured learning program is also needed to develop a data storytelling mindset and enhance the workforce’s data literacy. Even the MIT Management Education Executive program has a module on communicating data through storytelling.
By realizing the need for scaling data storytelling skills, I partnered with two experts — one leads IBM’s Data Science Profession worldwide, while the other is a Principal Data Scientist and Data Visualization Researcher— to devise the first data storytelling curriculum for IBM’s employees. We designed it as a set of three modules —associate, professional, and advanced— to offer everyone the choice of how deeply to get into the learning content. In fact, people don’t need to acquire the same skills: different roles call for different levels of knowledge to work with data. Not everyone must become a Machine Learning engineer or a data scientist! The intent of the program was to establish a baseline for data and AI skills from which everyone could start learning and progressing.
To teach how to craft data narratives, we assembled in a single program skills that usually are not taught together, ranging from data analysis with Python and machine learning fundamentals to data visualization design, visual storytelling, design thinking for data and AI, and AI explainability.
We laid out the learning program with the support of the IBM Data Science Profession and the IBM AI Skills Academy to also get executives involved in the data literacy effort and raise awareness of its importance. Since its launch in December 2021, the program has contributed to infusing a data-driven culture and has been attended by more than 1,700 people.
I see this program as a stepping stone for people to discover new opportunities and uplift themselves thanks to the data abilities they acquire that bring them to fulfilling their daily job more creatively, efficiently, and faster. By training people to use data as a rhetorical device, which comes with thinking critically and strategically about it, companies will nurture a new generation of leaders. By embodying both the analytical and rhetorical side of data, these new leaders will respond to arguments with clear and thoughtful communication by using data to back up hypotheses and challenge assumptions. These data-savvy humanistswill be the ones that will keep algorithms in check to preserve the governance and trustworthiness of enterprises’ models; they will be the future of organizations, the ones pushing them forward by transforming data and algorithms into great visions illuminating companies’ path towards a responsible digital transformation.