Designing with usage in mind

Visualizing data enables faster decision making. Built-in charts and graphs often provide good ways of showing data. However depending on the audience, environment, and unique circumstanes, it’s necessary to be creative.

The visualization below was originally conceived with the goal of showing a customer the potential that using Power BI had to change the way they run their operations. It needed to quickly convey to a CxO level audience the impact the technology stack could have, while also being useful in its own right.

The primary goal of the visual was to take enable a CxO level audience during a board room session to relate learnings from their investment in machine learning to strategies the organization has identified as critical to their success in market. These are deeply abstract concepts that could best be related using a visualization.

This interactive was designed to help CxOs understand and focus on their chosen strategies and machine learning generated recommendations

This interactive was designed to help CxOs understand and focus on their chosen strategies and machine learning generated recommendations

I worked with a team to create the concept and visualize through a number of iterations. While visually it makes an immediate statement (intentionally so – it speaks to uniqueness, customizability, unlimited potential), the visual concept started as a humble Sankey diagram.

I chose a Sankey diagram because it shows contribution of inputs through a set of abstract stages until they materialize as outputs. The problem here was to ground a discussion of machine learning recommendations in an understanding of core strategies meant to improve the success of parts of a business – in this case parts of an airline.

Just like in a Sankey, it’s not always clear what specific input resulted in what specific output. However, the flows and magnitude of the flows paint a picture that can tell a story and focus discussion, which was the goal here. Take a look at the work by the Pew Research Center telling the story of Democratic voters and the road to nominating Hillary Clinton. It illustrates my point very nicely.

However, a regular Sankey diagram would have been confusing for the CxO audience. It would not enable the board room to focus on just one part of their business when needed.

Worst of all, a traditional Sankey shows a flow of some quantity through a set of categorical stages. That quantity needs to be consistent – money, number of delegates, number of errors – doesn’t matter what, but it can’t morph from one stage to another. You can be talking dollars then change into facebook likes (hint: use scatter plot for that!).

In our design problem, the numbers did not represent a flow of one quantity. So I removed that aspect of the Sankey diagram and inserted an element similar to a network diagram with nodes representing each part of the business, and a clear grouping showing the busines as a whole.

This chart shows corporate strategies on the left with weighted lines indicating the importance of each strategy to each airline segment in the center. The right side of the chart shows critical success factors and the lines on the right side are weighted based on machine learning recommendations.

This chart shows corporate strategies on the left with weighted lines indicating the importance of each strategy to each airline segment in the center. The right side of the chart shows critical success factors and the lines on the right side are weighted based on machine learning recommendations.

During a discussion in a board room, it’s critical that participants don’t confuse which strategy applies to which part of the business. The interaction on the visual was designed to enable a view of the entire business and to focus down on just one part of the business. By clicking on Budget the visual shows only the data relevant to that part of the business.

Since the report is interactive, when a user select Budget, the data shown in other parts of the report focuses to show just this piece of data.

The interactivity of the visual was designed specifically for board room scenarios, where focusing on one part of the business at a time is critical for productive discussion, but also enabling an overall view of the data making sure nothing is obscured.

The interactivity of the visual was designed specifically for board room scenarios, where focusing on one part of the business at a time is critical for productive discussion, but also enabling an overall view of the data making sure nothing is obscured.

I would not recommend this visual to draw analytic inferences. However as an infographic that helps keep an audience engaged, drives clear understanding of very complex sets of associations, and shows the potential for a platform, the visual achieved it’s design goals. It continues to be a perennial favorite to show as part of Power BI demos helping customers understand the options for visualization they have available to them, including building custom visuals to meet their own unique needs.

Ideas in brief

  • Understanding requirements can lead to novel visualizations
  • Traditional visuals are not always what’s needed for a given situation
  • Customize interactivity to have the effect you desire
  • Use ideas from existing visuals and iterate towards your desired outcome

 

Notes

This project was a collaboration with John Ratajczak and Jonathan Howell on Design and Sachin Patney on implementation. You can view the visualization in the Airline Industry Solution in the Power BI Solutions gallery. In the process of designing this visual the team reviews many different visualization, including Chord diagrams; just goes to show that discussion and debate are needed to clarify ideas and get to the final solution.

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s