Skip to main content

The Art and Science of Action-Driven Visual Analytics

Headshot of article author Miranda Li

Last year I went to our CMO Chris Capossela’s talk called “What’s Great Data in Microsoft”. In this talk, he listed five of the most important characteristics of good data: self-describe, fresh, forward thinking, inclusive, and adopted. The last one – adopted – is what he emphasized the most, and he challenged us to think harder about turning data into business actions.

How many business actions have you provoked with data? What methodology and thought process do you use to achieve these successes? Please contribute your ideas below in the commenting field.

In my experience, there is a lot of art and science behind every action-driven visual analysis. From defining business problems, analyzing data, designing powerful visualizations, constructing impactful narratives, this is the space where business meets data, art meets science, emotion meets technology, and creativity meets rationality. To convince our users to act on something, you must tackle them all.

Below is a graph I created based on Steven Few’s Tapping the Power of Visual Perception, John Medina’s Brain Rules, and a range of cognitive psychology articles I read. As humans, we face large amount of data and information every day. To derive meaning and make sense of this world, we constantly scan the world around us and selecting what is important and what is not. Iconic memory is our first layer of filter, and we will not pass on this information for further processing (short-term memory) unless we have an interest in the subject. Actions? It is even harder. How likely are we willing to act on anything if they don’t touch our hearts, bring us benefits, or prevent us from losing something (long-term memory)?

clip_image002

In the world of big data, a visualization is merely a vehicle – a vehicle for us to create patterns, familiarity, and salience with data so that we can attract users’ attention and tap into their iconic memories, but to convince them to take actions, we must think deeper and tap into their short-term and long-term memories: Who is my audience? Why should they care? Will I make their jobs easier and help them create more impact?

With this framework in mind, let’s look at the two data visualization examples below and see which one is more effective? For illustration purpose, let’s assume that the user for this data visualization is a project manager at an IT Consulting Agency. Her performance is measured by the number of projects she does and how quickly she delivers solutions to her customers. To achieve high impact, she constantly looks for areas that hinder her effort or projects that drag down her performance.

Below are two different data visualizations that try to impact the same project manager. Please read and think about which one is better at serving her need and why? Will any of these visualizations provoke business actions and why?

clip_image004

clip_image006

I think the thought process behind the 2nd visualization is more apparent than the 1st one. Considering the project manager is concerned about project speed, the 2nd visualization has added a “Above Median” calculation to separate concerning projects from the ones that are going well, and then effectively used visualization tools to surface those items that need attention:

  • Added a multi-pane dot chart to display all projects and surface those concerning projects by using salience (red stands out against blue). You can download MAQ software’s dot chart from the custom visual gallery.
  • Added “Variance from mean” to the text table, use data bars and color (same red/blue as the dot charts) to highlight projects that are above means.
  • Arranged the two bar charts together at the right, made them both vertical bars, and aligned their categories so that we can compare effective between the two charts, e.g. simple project counts and simple projects time are aligned vertically.
  • Added a title to top of the tab to illustrate the focus of this visual
  • Added a KPI filter to allow users filter out non-actionable items and focus on the actionable ones.

The magic of action-driven visual analysis is never about the beauty of the chart, but rather the thought process that goes behind it: identify what is important for your audience, and then use visualization tools to surface what they care about.

To learn more about this framework, check out my recent webinar recording for Power BI customers and prospects. Have great ideas on how to drive actions with data? Join the discussions by using the commenting field below.