ds-blog-icon.png

THE DATA DOWNLOAD

As leaders in Enterprise Data Management and Business Intelligence, we pride ourselves on being able to deliver successfully on our projects as well as provide high-quality content to our readers.

Data Visualization Essentials: Telling the Right Story

Posted by Gordon Reynolds on Jun 2, 2016 9:28:27 AM

Effective visualizations support good data storytelling. They guide the audience through the data in a way that allows them to quickly understand and easily draw conclusions. Understanding the audience, taking the time to prepare the data, and choosing the most appropriate visualization type, ensures visualizations will deliver clear information that can then be acted upon. The sections below outline the essentials to create both compelling and actionable visualizations for any audience.

Prepare the Data

Raw data is rarely ready to be used by visualization tools, and proper data set up is a critical, yet often overlooked step. Although the process often requires a significant level of effort, taking the time to extract and structure the data provides the foundation for effective visualizations.

data visualization essentials.png

Using structured data organized hierarchically within clean levels, supports exploration and helps the user identify the pieces needed to tell the story.  It is also important to consider the different types of data that will be used to create the visualization. Some data can easily be aggregated to any level, while others cannot. For example, if sales representatives are counted by products sold, what happens in the case where one representative sells more than one product?  We now have the potential for counting the representative more than once, if we decide to roll up the data to a higher level.  The same is true when trying to calculate a percentage in the data, rather than taking in the numerator and denominator and allowing the visualization tool to do the work at the level we choose. 

During the data set-up and exploration process, maintain focus on the detail of the story to be told.  Aggregating data to the highest level to create a single graph or chart may serve an immediate purpose but may not support additional needs.  For example, a single graph for a company may show both a positive profit over the last five years and growth during each year. However, there will likely be additional questions, such as why it is growing, or how is each division performing. Considering these questions during the data preparation stage provides the foundation for moving to the next step, selecting the proper visualization.

Selecting the Proper Visualization

There are many different types of visualizations, each designed for a specific purpose depending on the communication goal. In the realm of Business Intelligence and Enterprise Data Management, effective visualizations will be much more sophisticated than the familiar basic graphs. However, the detailed visualizations used in our industry are almost always derivatives of those basics, and the core concepts still apply.

  • A bar chart is great for comparing multiple items based on a single attribute.
  • A line graph is more effective at showing time series and change over time, especially when comparing multiple items across time.
  • Pie and brownie charts are great at comparing items not only to each other but also as part of the whole.
  • Scatter plots are effective at showing the relationship between two different metrics.

Once the best visualization is identified, it is important to select the appropriate options to effectively tell the data’s story. This concept is easily understood when considering examples of simple bar charts and line graphs.

Example 1
example1.png

A bar chart will only tell the story correctly when proper axis values are used, since the size of the bar is how the audience quickly identifies differences.

  • Example 1 (above) shows two bar charts using the same data (profit by year), but each chart represents the data differently.
  • In the left chart, the axis starts at 385 instead of zero, making it appear as though profits are growing at a fairly fast pace.
  • In the right chart, the axis starts at zero, showing the gradual nature of annual profit while accurately portraying the performance.

Example 2
example2.png

Using the proper axes data type is also important when creating line graphs.

  • Example 2 (above) shows two line graphs using the same data (units sold).
  • In the left graph, unit count makes it difficult to determine any directional variance in the three products’ performance.
  • In the right graph, percentage change from prior year is used to show the three products’ performance, making it easier to see the directional trends relative to each product.

Effective Dashboards

Dashboards can be an incredibly effective way to tell a story, by displaying a large amount of information in a single view.  Dashboards are most successful when they are clearly laid out, show the optimal amount of data, and only display relevant data. They can tell several different stories clearly or can tell a single data story from start to finish.  For example, compare an executive dashboard to a dashboard for a single department.  The executive dashboard could include data from multiple departments or key performance indicators to provide a high level view of organizational operations. Conversely, a department dashboard offers a specific, detailed view, allowing you to identify the drivers of performance trends uncovered in the executive dashboard.

Telling Different Stories with a Single Dashboard

customer_segment.png

A single dashboard, if designed correctly, can tell multiple stories at once. Consider the example above, year over year profit is compared to the total order quantity for different customer segments. The consistent formatting and style leveraged allow us to quickly evaluate the information to make quick determinations of how each business segment is performing.

  • To tell several different stories clearly, a dashboard should:
    • Label sections so each story is clearly defined
    • Include sections to meet the needs of the intended audience
    • Contain only relevant information
    • Feature stand-alone visualizations for rapid assessment of the individual metrics displayed

Telling a Single Story with a Dashboard

If leveraged properly, you could also use a dashboard to tell an encompassing story. For example, consider the dashboards mentioned previously. However, instead of evaluating one metric, the dashboard would incorporate multiple metrics for a single segment (see below). This dashboard allows the audience to dive deeper and form a conclusion that supports an observation made from an executive dashboard. 

profit_by_state.png

A properly designed dashboard will contain different sections that are connected and will filter based on selections made, allowing us to make selections that will guide us toward the actionable conclusions we want to identify.

  • To effectively tell a single data story from start to finish, a dashboard needs to:
    • Provide easily understood metrics that communicate relevant points to the target audience
    • Communicate the relationship or progression of one section to the next, often starting at a higher level view of the data and then drilling down to show the data contributing to those totals
    • Contain only relevant information to support a conclusion
    • Leverage data for actionable insight and informed decision making

Wrap Up

Businesses rely on dashboards daily for decision making and performance insight, and every day new tools and best practices are developed.

While this article provides simple considerations, your project likely has its own unique requirements and demands. Regardless of the approach or tool(s) you leverage, the most important thing required is consistency. In order to create compelling visualizations, you must adhere and not deviate from the standards you have put in place.

 


Are you ready to discuss your project?

Let's chat about it, we look forward to helping and becoming your data partner. 

request-call-with-practice-lead

Topics: Business Intelligence, Blog

Written by Gordon Reynolds