Data Visualizations that Really Work

Data Visualizations that Really Work

The June, 2016 HRB article by Scott Berinato examines how to use charts and diagrams to express ideas and statistics. I agree with Anmol Garg, Tesla data scientist quoted in the article, “You can’t find anything looking at spreadsheets and querying databases. It has to be visual.”

Bernato says: “Convenient is a tempting replacement for good, but it will lead to charts that are merely adequate or, worse, ineffective.” He separates visualizations into four components: idea generation, idea illustration, visual discovery and everyday dataviz. Simple line, bar and pie charts are great for idea generation and illustration, but terrible for visual discovery and dataviz.

Visual discovery is the home of Six Sigma: testing hypotheses and data mining for patterns, trends and anomalies.

While managers do most of the everyday visualization using line, bar and pie charts, these charts are the dumbest charts around. They barely graduated from high school. Corporate dashboards are filled with them. These charts waste resources chasing the short-term rise and fall of key performance indicators (KPIs). Corporations need charts that went to college and took statistics–control charts, Pareto charts and histograms. Charts that can tell you when performance has truly shifted.

Berinato concludes: When we make a good chart, to see what couldn’t be seen before, it changes minds and causes action. Control charts, Pareto charts and histograms can make the invisible visible. Line, bar and pie charts cannot.