Let Your Data Do The Talking!

In a world of data, it is surprising that we often lack the ability to harness the power of that data. The goal of this article is to share my data storytelling process to unlock the power of your data! I will share tips for how to get to know your data, tell a compelling story with your data, and develop and implement effective visualizations that empower data driven decision making in your organization.  This article is just the foundation but should help you to get on the right track to let your data do the talking!

Know Your Data

When it comes to getting to know your data, there are three parts that you need to explore; the audience, the context, and the data. Getting to know your data will enable you to develop a clear and concise story or message for the stakeholders. 

Detail is key in understanding our audience. Ask yourself questions like: Who is the decision maker or key stakeholder? How does this person perceive my work? Do they know & trust me, or do I need to develop credibility?  Once you identify who the audience is, you need to determine what and how you will communicate with that audience. What do you want the audience to do based on the visualization or dashboard? Do you want to inspire them with an infographic, do you want to persuade them with a short case for action and supporting data, or do you want them to interact and take action on their own data insights? If you can’t identify an action for your audience, then reconsider if the data or visualization is really needed. Finally, how will the data be presented? Will you be presenting a static visualization in a report or email? Will you be giving the data story in a live presentation? Will you be developing an interactive dashboard for a team? Narrowing the focus of the audience will help us to select the most effective visualizations and data for our message and add enough detail for the delivery method to be successful.

You know that saying; “A picture is worth 1,000 words”? I believe that this can be true for our data visualizations if we take the time to understand the context! Understanding the background and context allows you to tell the full story. You can start by investigating when, where, why, and how the data was collected. What is the intended purpose of the data- was it collected for a specific reason and does that reason bias the data in any way? Identify if the data is collected before, during, and/or after an event and on what frequency. Connect this timeline to major events that could impact your data. For example, when looking at traffic data, annotating COVID lockdowns may be valuable to explain unusual data changes and impact the decisions your audience will make. Finally, how was your data collected? Data could be collected in written or virtual surveys, automated web data collection, sensor readings and many other formats. When you understand how the data was collected, you will be able to assess the quality, validity, and usability of the data. 

Now that you have taken time to know your audience and context, it is time to open up the data! Explore the raw data for fields, format, gaps, and potential errors. If the data is from a human source, notice where spelling mistakes or other human error may impact the data quality. Next, use a visualization tool to explore the data. Visualize trends, distributions, correlations, gaps, and other patterns in your data using various charts like histograms, scatter plots, and paretos. Finally, apply context to your exploratory analysis by considering the impacts of the data contexts on the patterns and trends you’ve identified. Remember that these visualizations are intended to help you, the developer, to see what is in the data set. These visualizations will not (usually) be a part of your final product. 

Tell The Story

Telling a story with data is a very powerful and logical process. You first need to start with identifying the right chart type for the data relationship that you are portraying. Data relationships include deviation, correlation, ranking, distribution, change over time, magnitude, part-to-whole, spatial, flow, and more! The Financial Times has released a fantastic resource, called Visual Vocabulary, that I use for all of my visualizations. Start by asking yourself what question you are answering with the data, and use that to help you define the data relationship. To answer “What region has the highest sales for the last quarter?” you will need to show a rank relationship. To answer “What product has the least seasonal variation in sales?” you will need to explore a change over time relationship. Now, take a look at the relationship and scan through the chart types in the Visual Vocabulary poster. Notice that the most simple or effective visualizations are at the top and some of the more complex visualizations are at the bottom. Choose a visualization that your audience will be able to relate to easily. 

Once you have chosen a chart type, you are ready to change all of the defaults on your chart! Really… I mean almost all of them. Default settings on most chart types add a lot of clutter that distracts the eye of the audience from the data and story you are presenting. Where possible, you will want to remove unnecessary borders, text, and data points. In the example below, I’ve taken the standard bar chart that is ordered alphabetically, reordered by height, and then removed unnecessary grids, axes and labels. 

Finally, we can use preattentive attributes to draw focus in our visualizations. Preattentive attributes are visual properties that we notice without using a conscious effort to do so. Some examples of these attributes are length, size, position, and color. I really like how this article explains the application of preattentive attributes in data visualization.

One word of precaution here; adding too many attributes can result in clutter, not clarity. In the example below, Image 1 uses color, size, and enclosure to pack as much information into the visual as possible. Image 2 on the other hand uses color to direct the attention of the reader to the point that is stated in the title of the chart. Notice how long it takes for you to digest the information in Image 1 as opposed to Image 2. The cognitive load of too many preattentive attributes becomes overwhelming and no conclusion is drawn from the first image. When it comes to preattentive attributes; less is more!

Implement

Once you have answered the questions being asked, it is time to bring it together. These principles can be applied in any format or method of communicating data, but for simplicity, I will focus on dashboard design. When implementing your visualizations in a dashboard, there are 3 A’s of design that I like to keep in mind; Affordances, Accessibility, and Aesthetics. 

Affordances are aspects that are inherent to the design that make it obvious how the product should be used. In data visualization, you can think of this as a chart axis or legend. Our audience knows exactly how to read that information because it is inherent to the design of a chart. We can also use affordances such as the size of text to draw our audience to a title, then subtitle, then labels simply by ordering the font size in the order you want your audience to read the information!

Accessibility is all about choosing design elements that are usable by people of a variety of backgrounds and technical skills (or the specific background and skills of your audience). I like to keep it simple here and use text to bring accessibility to my dashboards. If I’m designing an executive dashboard, I use minimal text, large font, and really focus on the most important elements of the data rather than excessive detail. I tend to hide the details in filters or backup sheets that the executive can drill into if needed, but the first page they see gives them a 5 minute digest on the key elements of their business. If I am delivering a static visualization, I give the BLUF (Bottom Line Up Front) in a descriptive title that they can then confirm with the data I present in the chart. When building a dashboard for a technical team, however, I may offer more detail, interaction, and supporting text for the user to digest the data that is available to them and draw their own conclusions. Another method that I like to use for some audiences is using a question for a title. The question that the chart is answering will guide the audience in what conclusion they can draw from the chart and then move on to the next. Accessibility is all about knowing your audience, and what their needs are.

Finally, aesthetics- making it pretty! You don’t need to be an artist or a designer to spend a few minutes designing the look and feel of the dashboard. If design isn’t your thing, here are my tips on what to focus on:

1. Alignment: Misaligned titles and axes are distracting for the eye. By simply bringing your visualizations into a logical alignment with one another and maintaining a standard font size and alignment for titles, you can free up your audience’s mind to focus on the data!

1. Color: START WITH GREY! I don’t know that any designer would ever give you this tip, but I think that it is the most effective way to implement color in a dashboard. Focus on the data story first, and then add color. Either give the color meaning, like representing a sales region, or other series throughout the dashboard, or use color to brand and theme a dashboard. If your dashboard has more than 4 colors on it, stop and rethink the purpose of color in your dashboard. In the example to the right, the #IronViz champion, Lisa, used color to tell the story of three artists who made breakthroughs in the music industry.  Our eyes can easily relate these 3 colors across the different visualizations to follow the artist’s story. 
2. White Space: Don’t be afraid of white space (or blank space between visualizations)! The padding or spacing between visualizations gives a less cluttered look and feel to your dashboard. It will allow your user to process information as they move on to the next visualization on the dashboard. 

Learn More!

This article only brushes the surface of knowing your data, telling the story, and implementing visualizations. Let me know what topics you'd be interested in me diving into more detail about! In the meantime, I’ve joined Jane Li in a 3 part series on Vizlib’s Appy Hour to talk about this process for building your data story. I encourage you to watch these episodes to see more details and practical applications of the topics covered in this article! 

Part 1: Know Your Data

Part 2: Tell The Story

Part 3: Implement

You can also follow me on Instagram to see more of my data stories!