How do you prove your data visualization is valuable?

There are 2 questions that you should always ask yourself when creating any data visualization. 1. What will the output be? This will typically either be a reoccurring report, a one-time presentation, or you could just be doing your own ad hoc analysis. Each of these require different strategies for how you set up your visualizations. 2. Who is your audience? A reoccurring report for your CEO should look considerably different than a reoccurring report for a group of analysts. A CEO will want to spend 5-10 seconds and have a good idea of what is going on. An analyst will want the ability to dig into root causes and significantly more granularity.

The proof of the pudding is in the eating." Drawing from this proverb I can infer that the true value or quality of the data can only be determined by experiencing or testing the results it generates from the decision that has been taken. In the context of proving results, it implies that the effectiveness or success of data visualization can only be established by observing the actual outcomes or results it produces. Simply put, actions speak louder than words, and tangible evidence is the best way to validate the effectiveness or value of data.

"Value" can mean so many things. Ultimately you want the data to tell a story and different people want to hear different stories. This doesn't mean the data changes but the way it is visualized can help different audiences to draw valuable insights based on their own perspectives and experiences.

Don't underestimate the power of metadata as well! Sometimes collecting data around the data can help understand why the data might be the way it is.

It can also be helpful to document notes about the data set during the exploratory analysis phase. Noting simple questions you asked about the data may help shape the narrative and output when presenting to the audience.

Data visualization is all about communication. It is getting data to your audience in a way that allows them to consume more in a shorter amount of time. The goal of any visualization report should be to make it easy for your audience to get the information that they need. By far, the biggest mistake I see in data visualization is the creator overloading the audience. Anything that does not provide a significant amount of additional value should be considered to be removed so the most important pieces of the report can shine. You are not done when you can’t fit anything else onto the page, you are done when you can’t take anything else off.

Data visualization is an iterative process and in my rush to insight I sometimes stop with what appears to be the first right answer. I find (and teach) that it is important to apply iterative questioning to data visualization - a cycle of asking, answering, and refining questions to help explore, analyze, and communicate the insights in data. Iterative questioning can take various forms, but because of my own tendency to "get lost in the weeds" as I dig into the data, I find it useful to intentionally "zoom out" from time to time to evaluate my visualization and ask if it answers my original question, or if - perhaps - I've found a better question for my goal.