Being an analyst with modern visualization technology is fun and enriching. Who doesn’t enjoy a stunning visual with useful and easily consumable insights? The reality is: even the most thoughtful authors have blind spots. Why? Because authors have significant amount of background knowledge that is not carried through via visualizations. In the absence of that, users relay on visual cues and the knowledge they learn from elsewhere to interpret what they see. This often cause communication gaps in data visualization, particularly in situations where report is authored by one person and used by many users from different teams, different backgrounds, and have different background knowledge and chart reading abilities.
Like here, a great visualization published by our Power BI enthusiast JaredK . The gallery title is Sales Scorecard: Where are we losing money. From this very informative title (I wish it was on the visualization though), I immediately see the intent of this visualization: help sales management to identify where we lose money. As I move down to the visual, my eyes are locked to where the author wants me to: The RED - Forms, Virginia, Indiana, Accident Reporting under Melonie Wiesner, and several other areas. By aligning intent with powerful visual cues, the author gets his key points across in less than 5 seconds. How can I not be a delighted customer?
But as I dive deeper, a few questions arose:
Are those sales numbers on the top left charts? Where are the labels?
The bottom left table must be profit? No labels either.
What are the big callout number on the right charts? Maybe year over year growth? Since there are current year and previous year number beside it.
These seems nitpicking. But require guessing in data visualization not only takes effort, but also can be dangerous. Add a few extra words or labels on the sub headers and callout numbers will eliminate the need for guessing and enhance user experience.
Here is another example. Can you see where are the authors’ blind spots?
What are the area charts and bar charts trying to measure? Judging from the shape of the trends, I assume the top two area charts are cumulative total sales, and the bottom two are monthly sales. Look another layer deeper, I realize all yellow colors are associated with services, and all white colors are associate with Hardware/Software. Again, a short header on top of each chart plus a good color legend would eliminate these guessing effort.
Where are the axis for the two area charts and the two bar charts? It’s hard to tell what the sales numbers are. Add axis and data labels would help.
The goal lines on the top are on the same level, but why are they different in numbers? Synchronize the axis would correct this.
The callout numbers at the bottom right are the same as the numbers inside the gauge charts. Are they redundant? Maybe remove the two callout numbers at the bottom left, and only keep the ones on the left?
This is by no means a criticism, but a reminder that blind spots are common for modern data visualization and consumption scenarios. The question is: how can we reduce it?
A simple solution I recommend is the FIVE SECOND TEST – Send your visualization to a coworker, give them five seconds to read your visualization, and ask them what they learned. If what they learned is largely different from what you want your audience to learn, then take your work back and try again. Another more thorough approach is to develop a check list: Does my title suggest my intent? Are my sub-headers informative? Is this the right chart? Have I labelled things correctly? How about interactions, tooltips, data format, and filters? The more thorough we are, the less head-scratching our users would need to be, and the more effective we are in getting our message delivered.