Designing experiences through data stories
Authors: Marion Hekeler, Mara Pometti, Ellice Heintze
Data is everywhere. Yet, it’s not always clear how to create meaningful content that drives action and decisions using data: the lack of a cohesive story connecting the data is the problem. At IBM, we have experimented with a novel approach that uses data storytelling as a catalyst for designing an experience enabling higher efficiency and providing more value for our users and customers. By collaborating with multiple teams across IBM we brought together designers, data journalists, and data scientists to renovate product design with a new and unique approach based on data storytelling.
While designing with data you most likely have come across the challenge of making data more consumable and trustworthy for users. Data appears intimidating and complicated to understand by non-experts because of its complex and tangled nature. However, we must remember that data is a by-product of our digital lives: although we don’t realize it, we generate data during almost every action of our daily life. This means that data is a human product before being converted into numbers and code. The real challenge we are called to address is how we might return to that human side of data using tangible and cohesive stories infused throughout products and experiences.
We believe that applying a narrative approach to data in the context of user experience and product design can eventually help users perceive the human side of the data they utilize and understand how it affects their jobs and lives. Dashboards, graphs, and visualizations do not necessarily tell the stories hidden in the data. Data visualization is a tool that we use to discover what the data is saying. Through visualizations, we reveal insights and separate findings—this is great for the purposes of a research process, but eventually we also have to connect the insights we find in the data to create an overarching narrative. Data stories provide a 360-view of the information a user consumes. These stories allow data analysts and other experts to feel more confident in their choices and more capable to solve their business challenges.
Data storytelling has recently become an official practice at IBM, as part of the AI Skills Academy — an IBM initiative that deepens technical skills in AI for key priority roles which focus on IBM’s Hybrid Cloud and AI strategy. Using a more story-driven strategy makes data and AI more approachable, experts can use this practice to enhance the user’s understanding of their tools. Data stories help people access their data easily and understand it better by providing context. This eventually generates more trust and results in better decision-making which drives concrete business outcomes and a more efficient experience.
Put data stories at the core of the experience
Crafting data as visual stories means working on the data in an attempt to clearly communicate insights in a consistent way. Connecting data in an overarching story helps put business questions into context and allows the business to move forward in an informed way. We always tie our designs back to a human-centered approach, focusing on our user . This ensures we remain focused on solving the correct problem, addressing our users’ needs and making their work more efficient. Everything goes back to creating a delightful end-to-end user experience. By doing that, we put the stories we find in the data at the core of the user experience.
By working in cross-functional teams and remaining open to experimentation of new design techniques, our team realized that data storytelling is a great tool to be used beyond just strategy and communication. It also offers an opportunity to transform the insights provided by our software into compelling stories. This involves a combination of three key practices: data design, data visualization, and a narrative.
But how do we put data storytelling to work in product design? Our team identified a list of best practices designers can use to incorporate more data storytelling into their practice.
If you are designing through the lens of data stories, make sure you hit 3 core dimensions:
- Trust & Explainability: Crafting a clear narrative and strong visuals helps users analyze and understand complex data. Insights are only as powerful as the data which fuels them, but inaccurate or flawed models can be equally catastrophic to analytics and AI initiatives. Make the right data available to the right people with trust and transparency. Provide context where the data lives, include a quality score, show where the data comes from, and be transparent about the owner of the data or who in the company used it previously. Such criteria helps a data consumer know their data better and decide if it’s trustworthy.
- Emotion & Memorability: Narratives and visuals are more memorable and evoke more emotions from the user. Emotion plays an essential role in helping our brains navigate the alternatives and arrive at a timely decision. When we package our data insights into a story, we build a bridge for the data to reach the emotional side of the brain. We combine visuals and micro-interactions together with a strong narrative to create more memorable moments.
- Persuasiveness: Communicating data in a humanized way plays an important role for initiating change and making efficient business decisions. By using stories to share the data, complex results and data analysis can be communicated in a simple way to stakeholders who are not data-affine. As mathematician John Allen Paulos observed, “In listening to stories we tend to suspend disbelief in order to be entertained, whereas in evaluating statistics we generally have an opposite inclination to suspend belief in order not to be beguiled.”
How to develop a data narrative in product design
Here are some steps to develop a data narrative in your product:
Start with research and audit your offering
Use data as a lens to frame human problems and reveal overlooked opportunities by fostering data-driven strategic thinking. Create an audit with user flows and identify data sources that should be embedded in a narrative.
- Can you communicate the user’s actions in a better way?
- Is there a high need for data visualization?
- Can you embed better explainability to make the technical AI process more humanized and transparent?
- Can you make data and insights more trustworthy?
- How can you communicate data in a meaningful and humanized way
Collaborate across disciplines
At IBM, we have people with all kinds of experience, skills, and expertise. Make sure you collaborate across multiple teams and leverage everyone’s superpowers. Bring designers, data journalists, and data scientists together to renovate product design with a new and unique approach to data.
Create a narrative
Information design is the foundation of data narratives: it uncovers the intricate connections that relate data to each other so to craft a story. IBM developed a bespoke framework that combines data and AI design with design thinking methodologies named Enterprise Design Thinking for Data and AI. Running a workshop using some of the tools of this framework, plus new ones specifically tailored to information design, helped us understand what data we wanted to focus on, and how to build a story with that data inside our products. By bringing the practices of data design into the mainstream practice of design thinking, we gathered all perspectives of the participants and aligned cross-functional teams on the narrative.
The process helped the team frame an overarching narrative encompassing all the data we surfaced and to understand how to visualize that story as multiple layers of information. Especially, the exercise of identifying and translating the information layers composing the story into visual channels, speed up the creation process and the data visualization exploration. In addition, by forcing us to think about data as the pillars of a story, the workshop fostered a data-driven mindset and made us approach the creation of new experiences from a totally new perspective.
Get inspired
It starts with inspiration, drawing big ideas, finding metaphors and analogies, and including each individual in the team. We take all these ideas, prototype them, test them with users, and then iterate until we get it right.
We create moodboards to find appropriate visualizations and micro-interactions that support our story. We look internally and externally for inspiration that makes our products engaging and helps us to communicate data in a meaningful and authenticway.
Data design guided by stories
Examples in IBM products
Example 1: Enable business users to create data stories
Data usually holds tremendous amounts of potential value, yet none of it can be utilized unless insights are uncovered and translated into actions or business outcomes. Complex analyses usually require some technical expertise which business users often do not have. They are asked to work closely with data scientists and analysts and are left waiting for more explanation. Business users commonly need to go back-and-forth to find the answers to any given business question.
We asked ourselves:
How can we help business users find relevant data that provides an answer to their business questions in an engaging and memorable way, and help them communicate their insights to others effectively?
Our goal was to give business users with less technical experience guidance about how to get relevant insights from their data. Then we wanted to show them how to communicate these insights to stakeholders that have not been involved in the analysis but who want to understand the essence of their findings to make smarter business decisions.
When typing in a search query or business question into a data platform, we usually return a list of data assets related to this query. Our idea was to move away from this list and to return to a visual representation of the data to give an answer to this business question.
Based on historical data and predictive models, the system picks 2–3 dimensions that are the most relevant to the business question and creates a visualization for them. For instance, if a business user wanted to understand the reasons for a drop in sales in a particular quarter, the system could generate a visualization that shows how sales (dimension 1) in this particular quarter (dimension 2) compares to sales in another quarter. This gives users insight into the extent of the drop in sales. This first data visualization serves as anchor point to explore data in an iterative and visual way, allowing business users to uncover the root cause of their business question.
Users can change the the visual representation of the data at any time to view it from another perspective. On the way to the root cause of the problem, they can also change and add dimensions that should be taken into account for the analysis. For instance, if they assume that customer complaints may have affected the drop in sales, they can include this factor in the visualization and it will show up as an additional dimension.
To build trust in the system generated insights, users can view which data assets have been used to generate the visualizations and assess them more in detail.
Moreover, we support business users in building up their own data story by providing them with building blocks of insights that they can combine in a meaningful way. They can create their own data story by dragging and dropping the insights that they consider to be most meaningful into a report. Then they can send this report directly to their manager.
Using this concept, we provide a guided experience for analyzing a business problem and supporting the end-to-end analysis process. Users can formulate a business question, apply hypotheses to the data and gradually uncover the problem. Then they can create a report that tells a story of the generated insights.
By pre-analyzing relevant data and converting them into meaningful visualizations, we go beyond just listing data related to a business question. Instead, we offer business users a visual entry point to start their exploration. We give them recommendations and cues they can adjust according to their needs. And we enable them to compile pre-generated pieces of insights into a comprehensive story so that they can make decisions and take actions based on these insights.
Example 2: Injecting AI transparency
In today’s world of digital transformation, companies are looking for AI capabilities to inform business decisions and automate tasks or processes. Intelligent systems can perform many of the manual tasks that employees currently handle, freeing their time to do higher value work and businesses to imagine new models. By offering more details on what is actually happening behind the scenes, we can remove this black box for users and develop trust between humans and the system. We explored the opportunity of communicating an automatic data matching process through visuals and embedded storytelling.
By educating with a simple visualization and clear narrative, we are able to reduce users’ uncertainty about the AI processes and use the loading time to explain a technical process in a humanized, simple way. Non-technical people are able to understand what is happening with their data — this in return, helps users build trust with the system and make informed decisions and actions based on trustworthy results.
Analyze and communicate data effectively
Visualizations are crucial and can open our eyes to insights we hadn’t noticed before. However, they need to be approached carefully. Something to keep in mind: we need to understand the data, pay attention to how information is displayed and know how to represent the data, otherwise we might be telling a biased story and charts could be misread. While working on a design, we might emphasize points that favor opinions or agendas just because of how we visualize things. We could easily mislead people by telling the wrong story with inaccurate visualization charts. Every time we construct a narrative with an alternative medium — whether it’s data storytelling, data visualization, traditional storytelling, writing, or even slide decks — we run this risk.
“A chart shows only what it shows and nothing else. The more the apparent message of a chart aligns with ideological beliefs, the more we should force ourselves to read it carefully, just because we are all prone to liking that chart beforehand and take it at face value.”
— Alberto Cairo, professor of visual journalism at the University of Miami, and author of the book, “How Charts Lie”.
Data literacy is a crucial aspect especially among designers. We need to be more informed and more careful. Information designers, data designers, and journalists need to know how to present the truth in the data and be transparent with their choices so that people are not misled and do not run into dark patterns.
Design principles
Lastly we recommend some guiding principles to follow:
- Embedded storytelling in the process elevates how we communicate data
- Consider what makes data and insights trustful
- Allow to take action at any point in time
- Reduce complexity, focus on simplicity
- Create a guided experience
- Add explainability and transparency to develop trust
- Consider flexibility & scalability for varying amount of data
- Use engaging visuals and micro-interactions to make the stories (insights) memorable and easier to communicate
- Design for everyone in a LOB without requiring technical skillset
Conclusion
Data storytelling represents an exciting, new field of expertise where art and science truly converge. It helps to deliver trusted, business-ready data to power future innovation, improve customer experiences and increase competitive advantage. By collaborating across multiple teams, product design can be renovated with a new and unique approach to data.
Special thanks
… to Ana Manrique and John Bailey for supporting and sponsoring this project and providing feedback
… to Erika Agostinelli for providing valuable user feedback and validating our concept along the way
… to Oliver Kauselmann and Andrew Smith for collaborating on conceptual designs and visualizations
… to Conrad Schmidt for technical expertise and prototyping
… to Justine Banbury for editing and providing feedback on this article
IBM’s statements regarding its plans, directions, and intent are subject to change or withdrawal without notice and at IBM’s sole discretion. Information regarding potential future products is intended to outline our general product direction and it should not be relied on in making a purchasing decision. The information mentioned regarding potential future products is not a commitment, promise, or legal obligation to deliver any material, code or functionality. Information about potential future products may not be incorporated into any contract. The development, release, and timing of any future features or functionality described for our products remains at our sole discretion.
Performance is based on measurements and projections using standard IBM benchmarks in a controlled environment. The actual throughput or performance that any user will experience will vary depending upon many factors, including considerations such as the amount of multiprogramming in the user’s job stream, the I/O configuration, the storage configuration, and the workload processed. Therefore, no assurance can be given that an individual user will achieve results similar to those stated here.
