Understanding Data Visually

If you have looked at many graphs and charts displaying metrics data you have no doubt scratched your head from time to time and asked what the chart is really trying to tell you, or why someone thought it important. The graphic portrayal of data should have one, and only one purpose; to make understanding of the data clear. Here are a few simple pointers:

Don’t measure and report something simply because you can. Measure and report what does or can make a difference.

Think about the timescale. Just because you can track it in real time doesn’t mean you should, or that it should be reported that way. If the variances over time are small, or if the opportunity to use it as a lever for positive change is small, then consider reporting the item on a quarterly basis.

Simplify for clarity. Too many charts confuse or hide what is really important because they include unimportant information and/or get too cute with the graphics. Graphics should be used minimally and to clarify, not to make the chart a piece of art.

Watch the value scale. Logarithmic scales are generally hard to interpret because we tend to take spatial relationships literally. If you have values with large spreads consider weighting them or annualizing them to normalize spread.

http://www.math.yorku.ca/SCS/Gallery/ is one of the best websites I’ve seen about visual presentation of metrics data. Take a tour and look at samples of good and bad metrics reporting. Then look at your own metrics reports and ask yourself if you are really telling the story the way it should be told.