The growing need for data storytelling
Data storytelling is the ability to communicate insights from data using narratives and visualizations. By 2025, Gartner predicts data stories will be the most widespread way of consuming analytics. Data stories were once the exclusive realm of professionals such as web designers, marketers, data journalists, and data scientists – but now we can access this magic with modern analytics tools. In fact, with information driving competitive advantage - it is an essential component of data literacy and communication.
We are all familiar with stories. We read novels, watch movies and are entertained with stories daily. It has been the historical tool for many cultures to make sense of the world, unite communities and communicate values and beliefs.
In an organisation, it is no different. Data storytelling is a process for communicating insights using data and visuals with a compelling narrative. It tailors insights to an audience – tells a story with numbers and promotes action. It can be a tool to foster better understanding, build communities and communicate corporate values and beliefs.
Data storytelling is exciting as a delivery means but also as a process. You can play with data like a forensics expert, while involving others in your journey.
Uncovering the story
Let us investigate the Enron Scandal – Wall Street’s most notorious accounting scandal. Enron was one of the largest companies in the United States – the darling of Wall Street, which collapsed overnight in 2001. The executives fooled regulators with fraudulent accounting practices and false holdings. Mountains of debt ended up hidden from investors and stakeholders.
The company’s email communications are available to the public. Listing the counts of Enron email communications between the senders and recipients in a data matrix and then visualising the information in a chord chart can yield interesting insights. The outcome is a visual web of communications – where we can determine the most frequent communicators and the gatekeepers for influential individuals.
Gathering and transforming the source email data, highlight these links in a visualisation and describing the relationships in an appropriate visualisation builds a story. The data story has actors in the form of communicators and recipients. We can ponder the relationships, the scandals and intrigue. We can zoom, filter and highlight relationships to discover insights and frame these to create a motion picture in the minds of an audience.
Download the data here: Enron Data
Jeff Dasovich has sent the most emails and has strong relationships. He may not have been involved with the scandal but electronically he had significant influence. Individuals like Tana Jones who received many emails may have been potential gatekeepers of information. These are insights that we can gather, combine and describe in a compelling way.
Storytelling in the workplace
In the workplace, great quantitative minds often struggle with communicating the value from their insights. Communicating complex concepts in simple terms is not an easy undertaking. Managers often have little patience for explanations on the rigour of mathematical models. They want to see the return of investment and business value.
But these bright minds can have game changing insights to deliver. The best sources of competitive advantage are often not obvious – and may be hidden in the complex data available. The best insights may similarly not be evident in the minds of the brightest employees – buried beneath their complex understanding. What is necessary is a methodology for any employee to easily create data stories and a culture in the organisation which fosters experimentation and the sharing of insights.
With the advent of self-service tools and democratisation of data within organisations, data-storytelling makes it possible to communicate insights from data and share these insights in a social context.
A developer may communicate system performance bottlenecks with the help of a bar chart; a business analyst may profile data and highlighting improvements to business processes; project managers may portray insights from project data to effectively gauge progress against budget and schedule. Roles across the enterprise can benefit in the same way.
The data storytelling process
The creation of data stories could be viewed as the iceberg's visible tip in a well-structured process. The data story is the tangible, visual outcome where insights are provided in a user-centric way - much like an app or web interface. Similarly, a step-by-step design process must occur.
Step 1: Know the audience
This step is critical. The type of audience will influence the type of data visualisations and the narrative language and form. Executives and managers may want high level strategic insights rather than deep technical details. An analytical audience may be interested in the justification for the insight and the data to support findings and theories. Some audiences are numbers oriented while others may be better supported by visuals.
Understanding the audience and involving them early in the creation of the data story will help ensure the insights are delivered seamlessly.
Step 2: Gather requirements
The next step is gathering requirements and understanding the business problem or questions. Well gathered requirements allows the data visualisations needed to manifest. Most developers that struggle to use the correct charts or graphs often do not have a design problem but rather a logical problem –not understanding the business question or domain well enough.
A popular requirement framework is the 5 W’s and 2 H’s. What, where, why, when, who, how and how much. Tackling user needs from all these perspectives helps refine the requirement and eventual solution.
These are typical questions:
- When are the best performing months for sales?
- Where are the best performing sales regions?
- How much sales occur in a day or month?
- Who are the best performing sales agents?
- How are customers shopping?
- What will the sales revenue be in a year?
- What do I need to do to optimise sales revenue?
As requirements are discovered, they are continuously evaluated and refined to discover additional requirements or hidden stakeholder intent that will benefit the client.
The requirements will guide the charts and visualisations used as well as the most important information to emphasise.
Step 3: Gather data
Data can be gathered when the business questions are defined. This process could involve using automated tools, scripting languages such as SQL, and enterprise data repositories such as applications, cubes, data lakes, and data warehouses.
Modern data tools can aid in profiling data and providing statistical summaries. This ensures that the data gathered is fit for purpose – aligned with the business question and of sufficient quality. Data wrangling skills are beneficial in transforming data to meet the intended purpose if necessary.
Step 4: Design the information flow
The information flow is the conceptual drill through to the required insights. It is the way we traverse the data to find the answers to the questions we have gathered. A mantra to achieve this is - High level first, zoom and filter and then details on demand. This provides a way to delve into insights in a gradual and planned way –answering questions at a high level and then diving deep to discover root causes.
The data storyteller can highlight a business context in a high-level view to create a setting for the discussion. For instance, you may describe the sudden drop in sales due to the Covid pandemic. A high level would give executives and big picture thinkers an excellent strategic view.
Zoom and filter
Zooming and filtering of charts provide tools to support arguments and hypotheses by highlighting exceptions and insights in charts and graphs. For instance, the storyteller may showcase well-performing branches or stores – the uptake of digital sales or a demographic of customers showing sales potential.
Details on demand
The resultant dataset – the detail on demand, is the final output of the visual exploration and the result of all the filters applied. We now have an action list – a set of customers or store items as examples – which can easily be communicated. This lowest level of information in a table or list can display the necessary detailed data for actioning.
View the data visualisation here: Nasdaq 100
Step 5: Create the visualisations
The correct visualisations help answer the business questions gathered in the requirements phase. They use form and design to provide quick understanding – visually drawing the user to patterns and trends. Typical categories for data visualisations include comparison, relationship, composition and distribution.
Comparison charts are used to compare one or more datasets. They can compare items or show differences over time.
Relationship charts are used to show a connection or correlation between two or more variables.
Composition charts are used to display parts of a whole that change over time.
Distribution charts are used to show how variables are distributed over time, helping identify outliers and trends.
Gestalt as a magic wand for design
Design principles help the data storyteller emphasize insights. Charts and graphs are visually modified to ensure ease of perception and speed of insight.
When narrating and presenting – highlighting elements in a chart or graph makes the experience dynamic, a static data visualisation now becomes a motion picture, and the storyteller can guide the audience to conclusions.
Gestalt principles are the visual magic wands. A human mind is a pattern-matching machine, and Gestalt aids our minds in finding the important patterns and images. This is useful for highlighting insights in charts and improving perception speed. Tools include using contrast, grouping, connecting and enclosing similar data points - highlighting chart insights and creating a visual hierarchy.
A treemap, for example, utilises several Gestalt principles. Colour may depict categories or rates of information change, which makes it popular with rapidly changing stock market information. The rectangle size shows a hierarchy with larger items in the top left and smaller items in the bottom right. The design has a sense of orderliness and seamlessness which draws the eye of the viewer to the most crucial insights – the biggest sales or the most significant risks for instance.
The chart on the left makes better use of the principles of contrast, hierarchy and orderliness. The bars are arranged orderly from the highest value to the lowest value. This is displayed in the order in which we generally view items - from left to right and improves visual perception.
Step 6: Create the narrative
The final step is to create the narrative for the data story. This structural framework links insights and visuals to a call to action. We typically use verbal or textual means to tell a narrative. You may have a speech to explain the charts or textual annotations to provide context to your information flow.
The storyline, plot, characters and settings are the ingredients for an exciting story. A typical narrative consists of a setup, rising action, climax and resolution – where tension builds the audience to a conclusion.
Source: Calvin and Hobbs
This involves highlighting the problem and setting the scene. Think of the Netflix series you have watched where the first episode consists of getting to know the leading players and the story’s context. The high level of our information flow is a great view to set the scene. We introduce the audience to the background of our story, the history and the main subjects.
The rising action is the series of events that increase tension and lead to a conclusion. We use this stage to allow the audience to discover insights. This involves the exploration of data using filters and slicers to drill down information and highlight insights. We introduce conflict in the forms of problems, opportunities and challenges. By gradually revealing insights, the audience is kept in suspense, much like the gradual uncovering of a plot in a riveting series that ends on a cliff-hanger.
Tension is built by gradually presenting possible resolutions to these challenges or opportunities until a climax is reached. The climax is the ultimate point of tension – where one must make a decision. Is the venture profitable? Do we continue and expand?
We tie together all the insights demonstrated, which leads to the final resolution.
The resolution is the further action to be made following the conclusion. The audience has an idea now of the expected outcomes. We may, for instance, decide to implement organisational change or an action plan based on the findings in the data story. This is our call to action.
Data storytelling is powerful and will become more relevant as we consume more complex data. As we practice this art, it can help us become more innovative and even foster a culture of innovation around us – where stories can be passed on an expanded.
However, the data storyteller is an emergent role that is still being debated and defined. Uncovering insights cannot provide value if these findings are not shared. Knowledge is useless without application. For the time being, let there be a data storyteller in all of us.
* Connect with Lyle on LinkedIn.
Watch a recording of our DVT Insights Webinar held on 5 May 2022 where Lyle Petersen talks about data storytelling. Click Here