Columns about bars

Comparing usage of Datawrapper chart types

It’s the last week of the year! We from Datawrapper are looking back on a year of two new hires and one returnee (we’re five people now), chart foldersa redesigned landing pagea new home for our blog (you’re on it right now!), lots of bug fixes, and four new chart types. If you want to know what the next year will bring for us, head over to Gregor’s road map!

Apropos chart types: Last week, Gregor and David pulled some numbers to see how often different chart types have been used since the start of Datawrapper’s existence a bit over five years ago. We decided to share these numbers with you:

You can clearly see when chart types were introduced. We started with the basics in 2012: tables plus bar, line, pie & column charts. Maps joined in 2014. This year, we released arrow, range, area and scatter plots.

Here are some observations that surprised us:

  1. Tables became big. We don’t know what it is with these tables. We offered them from the beginning, but as opposed to bar and column charts, they grew closer to our users’ hearts every year. In 2017, one out of six charts was a table – as many as the classic bar chart and column chart. Do we have an explanation? We have no clue. Do you have one? Let us know in the comments!
  2. A steady quarter of published Datawrapper charts are bar charts – and this never changed. Last year, we introduced new kinds of bar charts (stacked, split & bullet) and we’re happy to see them replacing some of the shares of the classic bar charts. But in total, 25% of charts are still some sort of bar chart. As we can see at the bottom of the chart, the column chart had a different fate: In the beginning it was seen more frequently than the bar chart, but by now the column chart is rarer in the wild than the bar chart.
  3. Fewer pie charts. Our users have built fewer and fewer pie charts for three years now, even in absolute numbers (they’ve created more donut charts in absolute numbers, though).

Chart choices

A chart about charts! I choose a simple stacked column chart for this data. Why? Because we have a ton of variables that we want to show (12 chart types!) and therefore need a chart type that offers us enough space to do so. A stacked bar chart doesn’t do that. It’s limited by the width of our website or device. A stacked column chart, however, can be as long as we want it to be. When showing many variables on a mobile device, a stacked column chart works better than a stacked bar chart.

Like with an area chart, one of the most important decisions we can make when stacking variables is their order: It’s not just a great way to lead our reader’s eye, like colour is. The lengths of bars that are placed on the same baseline are also easier to compare with each other than the length of the bars that float in the middle of a stacked bar chart.

I want to put the emphasis on our two most successful chart types (bars and columns) and therefore place them on the top and the bottom of the chart. But I also want readers to see that tables, maps, and our new chart types catch up. I decided to place them in the middle of the chart – to get the visual impression of them “breaking apart” the old charts. Even if these bars don’t have a common baseline now. I favored the visual metaphor over easier reading.

New Year’s Eve is just around the corner. Have a great night! We’ll see each other in 2018.