Data Visualizations with Purpose
Guidelines for creating meaningful data visualizations that drive outcomes.
As you prepare to create data visualizations, ask yourself these questions to create a compelling narrative.
What questions am I answering?
Data visualization goals are often tied to two main business objectives: communication and exploration.
Communication data visualization objectives focus on informing or enlightening the audience, by providing data or statistics on topics such as workflows and reporting. An example is to show revenue growth year over year.
Exploration data visualization objectives focus on prototyping, iterating, interacting and automating to answer strategic questions such as ‘why are revenues in this sector falling this year?’ Explorative data visualizations require a hypothesis (and therefore a null hypothesis) for testing and identifying the patterns, trends, and anomalies that may emerge.
What is my goal? Do I need a graph or chart?
Seeing your data represented through a polished visual can be highly rewarding, and Scott Berinato wrote of the temptation to ‘click and viz’, or the instant gratification that is available to quickly create a graph through business intelligence (BI) tools. Berinato suggests the guidelines below in ‘Visualizations That Really Work’ when navigating the connectivity between business goals and data visualization.
Which chart should I use?
Consider the variables that you are analyzing and select your graph according to the detail below.
Quantitative
- Continuous: Scale of infinite values. Statistics include mean, median, distribution, range, standard deviation. Examples of continuous data are Temperature (weather), Stock price
- Discrete: Count of characteristics, such as a result, item or event. Includes sum, count, mean, sum and standard deviation descriptive statistics.
Qualitative
- Categorical: Data with no natural order and nominal in nature. Examples include names, companies, departments and countries.
- Binary: (2) categories are compared, such as yes/no survey responses.
- Ordinal: At least (3) values with a natural order, such as a survey scale range (disagree, agree, strongly agree).
How could I design my graph?
Include all of the fundamental elements of a chart for your viewer. Your audience should be able to interpret the data represented on your graph in seconds, and simplicity is key. A few tips below:
- Colors: Use less than 6 colors in your graphs whenever possible. Viz Palette offers an excellent tool to assemble a color palette without color conflict, or distinct colors that prevent confusion for the audience when used in close proximity. Google Arts & Culture is another tool available for color inspiration, if you are looking for a place to start. Only use color gradients when comparing continuous variables where there is an incremental change in the numerical data (and therefore the graph colors).
- Avoid busy visualizations such as word clouds, and only use additional graph elements that add critical background for your story.
- Fundamental Elements: Include a title, include a legend, label both the X and Y axis (when applicable), include a legend. See image below.
