A rule of thumb that I follow for generating data visualizations: One story = one graph
The best visualizations are extremely simple and easy to read: e.g. a line or bar chart that tells you exactly what you care about
If you are struggling to figure out what to visualize, zoom out and ask yourself: what story are you trying to tell? Once you have clarity on that, figure out the simplest way to illustrate this.
If you have multiple stories to tell, make multiple graphs :)
Some made up stories and solutions:
Total engagement hours steadily went down over this year = You want a line graph of engagement over the year, and possibly you want to smooth your data out to show the trend line: e.g. graph the rolling 7 day average over time, or include a trend line.
Engagement really spiked on May 1, 2025 = You want a line graph of engagement per day, zoomed out far enough to show how it’s changed over time, and maybe add a labelled vertical line on May 1
Engagement this giving season is much stronger than last year = You want to plot two lines: engagement per day in 2025 over giving season, plus the equivalent engagement per day in 2024. Here, the comparison is the story you want to tell, so you want to make sure your 2025 and 2024 data are apples to apples.
Sharing communication advice a few colleagues have found helpful.
A rule of thumb that I follow for generating data visualizations: One story = one graph
The best visualizations are extremely simple and easy to read: e.g. a line or bar chart that tells you exactly what you care about
If you are struggling to figure out what to visualize, zoom out and ask yourself: what story are you trying to tell? Once you have clarity on that, figure out the simplest way to illustrate this.
If you have multiple stories to tell, make multiple graphs :)
Some made up stories and solutions:
Total engagement hours steadily went down over this year = You want a line graph of engagement over the year, and possibly you want to smooth your data out to show the trend line: e.g. graph the rolling 7 day average over time, or include a trend line.
Engagement really spiked on May 1, 2025 = You want a line graph of engagement per day, zoomed out far enough to show how it’s changed over time, and maybe add a labelled vertical line on May 1
Engagement this giving season is much stronger than last year = You want to plot two lines: engagement per day in 2025 over giving season, plus the equivalent engagement per day in 2024. Here, the comparison is the story you want to tell, so you want to make sure your 2025 and 2024 data are apples to apples.
Sharing communication advice a few colleagues have found helpful.
Great rule of thumb :) I’m sometimes knee-deep in chartmaking before I realise I don’t actually know exactly what I want to communicate.
Tangentially reminded me of Eugene Wei’s suggestion to “remove the legend”, in an essay that also attempted to illustrate how to implement Ed Tufte’s advice from his cult bestseller The Visual Display of Quantitative Information.
I’d also like to signal-boost the excellent chart guides from storytelling with data.