Data Storytelling: Why Visualization is only Half the Story

By taking a wider view of data storytelling, you can provide stakeholders with the big picture in a way that’s relevant and engaging.
Posted September 16, 2016

Guest Author

Data storytelling is everywhere – on people’s minds, in workshops and in analysts’ blogs. It’s so unavoidable, it makes you wonder: Is it a helpful way to use data or just another shiny new thing to feed the hype machine? We look at the more popular approach to data storytelling and discover how a broader view yields big dividends.

You Can’t Describe What You Don’t Define

The many points of view about data storytelling can make discussing it very confusing. Grouping the many definitions into several categories helps a bit:  

The use of data visualization software to engage and persuade decision makers to take action.

A translation of data analytics results, which take business leaders along the “last mile” of understanding from data to taking action.

A new and useful skill, which most data analysts lack.

A modern variation of traditional storytelling, which includes a "hook" to draw in audience members, themes, the use of emotion and a beginning, middle and end.

Taken individually, each provides limited ideas of data storytelling capabilities. Taken together, they provide a more revealing and useful definition of a new and powerful tool.

Stories That Are Accurate, Complete and Engaging

Most definitions of data storytelling involve telling compelling stories based on data. But recently, public discussions emphasize, even overemphasize the role of visuals and visualization tools.

This approach assumes that the way to make a data-centered story relatable and memorable is limited to using visualizations. Without doubt, this visual-centered approach makes data stories engaging.  But it’s no substitute for a complete and clear idea of what the data is saying – and to gain this understanding, one must look beyond visualization.

Understand the Story before Telling It

What many people who passively consume data are often not aware of - but almost everyone who analyzes data is – is that the bulk of the time dedicated to gleaning insights from data is typically devoted to data preparation. Visualizing the results of data analysis is the end result: but it is the final step in a multi-layered process that includes gathering the raw data, cleaning it, and modeling the data so that it can best serve the business’s needs and help answers ad-hoc questions from business executives.

This is especially true in today’s age of growing data complexity, in which organizations find themselves working with increasingly large datasets from many highly disparate sources. “Telling a story” about this type of data would generally require a lot of preliminary work to prepare the data before you go on to telling that story in a visual form.

In other words – before you start telling a story with data, you should do your best to first understand what that story is. The basic elements of this story would include:

Identifying the main characters: which are the data elements at play (databases, cloud systems, flat files, etc.)? Where are they located? How do you access them?

Understanding the relationships: Can you find a connection between these datasets? Which ones might logically affect others and how would you model these relationships (key tables, star schemas, etc.)?

Getting everything to speak the same language: How can you create mash-ups between disparate datasets that might come in wildly different formats, shapes and sizes? What tools would you need in order to clean the data and make structured, semi-structured and unstructured sources capable of communicating with each other using a single query language or API?

Once you have answered these questions and found suitable data analytics tools to help you build out the story, you can then go on to tell it in a visual form that will captivate your audience.

Tips for Telling Data Stories that Inspire Action

In the haste to make analytics results easy to understand and use, it’s possible to tell a story that’s inaccurate or incomplete.

Successful data storytelling uses all available tools to persuade its audience. It also balances the Big Picture references that ground the narrative with enough detail (data) to make the story believable and relevant.

A more complete definition of data storytelling—one that addresses all the definitions on our list above—will provide better results than a limited view. Here are ways you can guarantee a balanced, accurate and complete data story:

1. Start with high-quality data. Your story is probably built on analytics results of many types and sources of data. Before analysis begins, you’ll need careful data preparation to ensure accurate results. Theses time-consuming tasks can be performed manually by analysts or automatically by software.

2. Decide if there’s a story. Before you get into the details of storytelling, first decide if there’s a story to tell. Have someone familiar with the data help you find a narrative structure that will support and guide the story. If that’s impossible, look for another story elsewhere.

3. Remember your audience. Make the narrative and visualizations support the knowledge level and types of information your audience already has. This includes the different types of evidence that different stakeholder will expect and trust. Your story must address what audience members considers most relevant.

4. Find the compelling narrative. Have your data analyst describe the facts and how they connect to each other. Then, it’s up to business analysts to connect the facts with the most meaningful topics and business goals of the audience.

5. Collaborate, collaborate, collaborate. You have probably noticed that business and data analysts have important roles in this process. Make sure to wrangle them as well marketers and other experts for data storytelling projects. Then, encourage everyone to work and play well with each other.

6. Keep your finger off the scales. It’s easy to let visuals compromise data without anyone knowing it. Visuals might make some data seem more important than it really is in the story. There are many ways to combat this problem, including labeling to avoid ambiguity, providing graphic dimensions that match data dimensions and using standardized units.

7. Tell the complete story. What you leave out is as important as what is left in. Missing, outliers and out-of-range values; arbitrary time ranges and capped ranges can tell a different story than what you want to tell.

A More Powerful Approach to Persuasion  

By taking a wider view of data storytelling, you can provide stakeholders with the business Big Picture in a way that’s relevant, engaging and easy to understand. Bright visuals might be good for eye-candy, but often won’t suffice when you want to create a sensible and actionable story from a mess of scattered data. Once you have your story straight, use it to connect analytics results with specific decisions and business goals. Just think of it as another powerful and useful way to transform raw data into business actions.

About the Author

Ilan Hertz is Head of Digital Marketing at Sisense

Tags: visualization, big data, data analytics platform

0 Comments (click to add your comment)
Comment and Contribute


(Maximum characters: 1200). You have characters left.