There is an undercurrent on LinkedIn that has no idea what data visualization is and keeps repeating the mantra:
“bars for comparing values, lines for time series, pies for proportions.”
And everything must look familiar, at any cost. This ignorance gave us Power BI's sad visual layer of Power BI. Let’s talk about it, then.
First things first: we are free to do whatever we want when translating numbers into a visual representation (with a few basic caveats).
In a business context
In a business context, where (in theory) people prefer more effective charts, we need:
- Data, and a good understanding of it, obviously;
- Clarity about the business question the chart is meant to answer;
- The set of visual objects and their properties that we can use to answer effectively (the charting engine we use);
- The context (audience profile, distribution medium, etc.).
Good news: in a low graphicacy (graphic literacy) context, we don’t actually need to worry too much about familiarity (slight exaggeration). What we should consider are things like:
- How effective is the chart?
- How do users evaluate it?
- Do they recognize is as a significant improvement (insights, time saved)?
- Do they think it is worth the effort to make it a familiar chart?
- Can it be done easily in Excel? 😀 (that is, can it be integrated into existing processes and skills without causing major disruption?)
Excellent but less familiar examples
Here are three examples of effective charts that many people are not familiar with, and that don’t fit the “basic and familiar” chart narrative:
- Small multiples (to easily compare or monitor multiple entities);
- Cycle charts (to go beyond seasonality);
- The whole family of dot plots (far more effective than bar charts for two or more series).
What should we do?
So, what should we do, when in business context?
- Clarify the most common business questions/needs;
- Decide on the best way to answer them;
- Standardize as much as possible and, if feasible, define default settings in the chart library;
- Write and publish internally a style guide;
- Keep it a living document, revising and updating it as needed;
- Ensure the core ideas are followed, while leaving room for people to experiment with new approaches.
(Shameless plug: this is exactly what I help organizations do in my two-day, in-person data visualization course. So if you’re in Portugal, get in touch.)
Conclusion
In short: we need to improve data literacy and visual literacy within the organization. Don't change for the sake of change, but don't assume that a handful of old chart types can answer today’s complex and subtle questions just because they are familiar.
(Oh, and Excel can create far more sophisticated charts than you might think. I shared the example above previously in a comparison between bar charts and dot plots. And notice how the dot plot is much more effective than the bar chart.)
