Since my wife and I have moved to Calgary, Alberta, Canada, I've had some time to think about data visualization before I start a new role in a few weeks. Being that we moved to a city and country that start with the letter "c," I thought it was apropos to think about how I would describe good data visualization to a mentee. Before I list my wonderful list, keep in mind there are a plethora of material available on the web or at your library that go into significant detail about "good" dashboard design techniques (just google Edward Tufte, Stephen Few, or even the book "Art+Data" from Decisive Data). This is not a replacement of that material; merely, my own opinion and thoughts around this very subjective field. As cliche as it is, data visualization truly is both an art and a science. With that disclaimer out of the way, let's go on to the list!
Clarity. It's simple, yet a step easily missed. Be clear in what you are saying with your visualization. What question does it answer? What value does it provide? So many people will put together a dashboard, and after looking at all the pretty pictures, your stakeholders are left asking "So what?" Be clear with your message, and what the purpose is with your dashboard. Clarity goes beyond whether or not it's providing answers; it also impacts little things like font, spacing, etc. Is the dashboard physically easy to read? Too much chart junk? Clarity could also mean accessibility - if people aren't able to even find your dashboard, then it definitely isn't providing value.
Consistency. This might be more of a nit-pick, but having a level of consistency really adds polish a level of professionalism to your dashboards. The example that stands out the most for me is color. If you have a bar chart that colors the bars a different color, then that color needs to mean the same thing across the entire dashboard (or even dashboard tabs, for my fellow Tableau #datafam out there!). If a bar is red, is that good or bad? If you see red elsewhere, does it mean the same thing as before? Sometime we use color to identify values of attributes; other times, we use it to indicate a direction of a metric (think of the woefully overused stoplight color scheme). Regardless of HOW it's being used, color should be consistent. But, consistency doesn't stop at color! Having consistent font, sizing, spacing, even headers/footers, all matter. Back before modern BI platforms were common place, many of us in the BI space might have had to put together some sort of year end deck for senior level leadership; oftentimes, multiple teams provided the input to this deck. As a stakeholder, it's frustrating to go slide to slide, and the headers are all different, the colors are different, etc. The end deck shouldn't have looked like it came from 20 people (even if it did!) - it would have been better if the end user couldn't tell if it was built by 1 person or 20 people.
Context. This is probably the hardest part to bring to a dashboard, especially in a real-world business setting. But data is just data without context. In other words, it's not very useful. What if i see a KPI that has a score of 98%? Is that better or worse than expected? How does that compare to other business units? Or compared to the past 3 months? 12 months? Providing context helps paint a picture that shows our stakeholders what is going on. It helps the dashboard answer questions, rather than generate questions. Like photography, it helps frame the data so stakeholders don't get lost. Context helps drive a conversation and should help point out actions and decisions. Without it, you might as well be throwing darts.
If you made it this far reading, congratulations! One "C" I almost added here was Charm, courtesy of the folks over at Decisive Data from the book "Art+Data." However, I left it out because I think in a real-world business setting, it's trivial and takes away time from providing insights to our stakeholders. I 100% agree that people need to WANT to see your dashboard (a perfectly designed dashboard that no one uses is non-value), and I think when you look at some of the amazing dashboards people put together on Tableau Public are truly inspiring to me. That being said, it's difficult to do in a business setting, and ultimately I don't want people using their time to learn how to Photoshop an image to add to a dashboard. If you have the time, then go for it by all means! But, realistically, I think most of us are already short on time.
So, there you have it. Clarity, consistency, and context. I think if you can provide these 3 things to your dashboard, you're 95% on your way to a great story with data. This doesn't mean to say these are the only things to worry about - far from it - but, it's a good starting point especially for those new to the BI space. Do you agree? What am I missing? Would love to hear your thoughts!
Wishing my network all the best from Calgary!
