5typos.net
Read the chart out loud
When dealing with large datasets1 remember to tell yourself the story of the resulting chart.
Most of us usually create charts with some sort of agenda. We kinda know what we want to show, and therefore aren’t surprised with the chart if it fits our expectations.
The problem is that good data organized incorrectly can still look right. The most painless way I’ve found to try to catch these issues is taking a step back and telling a story of what the data is showing without thinking about your slide title. Just really read the data calmly, and you will likely catch a surprise or two.
Thankfully I’ve avoided a few charts with volume numbers until December 2017 (US vs world date formate), 1000x sales numbers (coma vs period thousand’s separator), and my favorite: 70 weeks per year (careful when how you use the DATE() formula).
Anything that requires you to scroll down I’ll consider large. If all the data is viewable, it’s easier to keep a mental model of it.↩︎