Simple on the surface, the faithful bar chart is often one of the first introductions we get to dataviz, but with its simplicity there comes a lot of potential to make incorrect design decisions that impact your chart’s effectiveness. For this reason we want to help by offering four points to consider when you’re next creating a bar chart to ensure the story within the numbers is clearly communicated.
Four checks to ensure your bar charts don’t fall short:
By having a basic understanding of the data, you can confidently make the correct display choice. First things first: a bar chart’s purpose is to compare categories.
So, if this is your intention, your next step is to plot the correct variable onto the correct axis.
With bar and column charts this means that your category – or your independent variable - should be your categorical axis, and your dependent variable on the height of your chart that the bars rise to – or the numerical axis. For example, for a chart communicating ‘fruits eaten in the office today’ – fruits such as apple, banana, peach are your categories or independent variables along the base of your chart, and the amounts such as 1,2 3, 4 are dependent variables along the height.
Watch out! If you are communicating distribution, rather than comparison, you will be a creating a histogram. Despite similar appearances, a histogram is NOT a bar chart. The data and the crafting of your chart need to be approached differently. Using The Data Visualisation Catalogue is also a helpful resource, as it lists out visualization goals, and suggests presentation options and charts you should consider to show the data.
If there is one thing we can stress, many feel that bar charts are ‘too basic’, often thinking that there must be a ‘better way’ to lay out the data. But the fact is that lots of data sets that show comparison are best represented in a humble bar chart. See what we mean below, the bar chart is the clearest option to present the story.
Ok, so you’ve ascertained that you’re communicating comparison and we’ve convinced you not to avoid using a bar chart. Your next design decision is if you should lay the chart out vertically – as a column chart - or horizontally – as a bar chart. This is more than just personal preference, and, sorry to say, but the different orientations should not be used interchangeably.
Thankfully there is a simple way to move forward, by following some simple questions. And in the name of visualization, instead of explaining too much with words, we have made you a lovely flowchart to follow.
You'll see that if you have less than 10 categories you have the choice between a bar or column, in most cases a bar chart (horizontal display) is the better option here, but depending on your graph it may be better as a column, so we'll leave this decision up to you!
A chart’s power to clearly and effectively convey information is often found in its simplicity. Restraint is what sets the best dataviz designers apart from the rest, with the beauty lying in insights shining through for the reader to discover.
Knowing how much narrative to include is certainly an art. We’re sorry it's not something we can supply you with another flowchart to provide the outcome! Instead, it takes practice to know when to restrain from explaining and marking out everything. Our biggest tip here would be not to underestimate the power of a useful and informative title to frame and guide the reader through the results and to really work on only including items (both graphic and text) which benefit the message you are conveying.
Also start consuming data visuals yourself. When you start to build up your own dataviz literacy, you will start to see what content is valuable and what is redundant.
One chart per message
Be aware that one data set can provide many different stories, so if you are needing to present multiple insights, don’t try and squeeze them all into one chart. Tell a logical data story by guiding your readers through a series of charts and key messages in order to have them understand. Think of it like throwing a tennis ball – if you throw one ball the chances of the receiver catching it is far greater than throwing, 2, 3 or 4 balls (which would be impossible!). Remember who the target audience is and what they need to take away.
Further considerations in order to remove ‘chart junk’ are:
Labelling items cleanly on the chart is far more beneficial to the reader, rather than including an elaborate legend that they need to go back and forth translating against the visual. Another benefit to direct labelling is that you can often remove your numerical axis.
I know, telling a designer this seems like we are preaching to the choir, but it's surprising how easy it is to lose a sense of your expertise when the word “graph” enters your brief. This is a tricky one, again there are no hard and fast rules to follow, it comes with practice, getting feedback and rounds of reviews.
Don’t be scared of white space! Allow the insights room to breath and tell that story that was hidden in the columns and rows of the spreadsheet. We mentioned it already, but it's important so it permits a repeat: make sure every element you have included on your chart has a purpose.
NEVER do this
No matter how tempted you might be, or how much pressure you are getting to ‘jazz-up’ a chart for next week’s board presentation, there’s one thing that will destroy your dataviz credibility.
Plain and simple, do not add 3D effects to your chart! It distorts the presentation and makes understanding the chart incredibly difficult.
And if you are still in doubt about if you have simplified you chart enough, check out this handy little gif by Darkhorse Analytics.
Not sure if you picked up on it, but there was a major theme missing from the point on simplicity above - some may say it's the most important - and that’s why it has its very own section. The topic of color could license its own blog post, actually even a series of posts, to cover the topic of respectfully - it’s a biggy. With lots available to read on the full topic, this tip barely scratches the surface.
But as a start point, consider this piece of advice: gray is your best friend.
We know there are so many clever ways to incorporate colorways from every which spectrum, they are just a click or two away and it's oh-so-tempting to start applying your application skills to your chart. But please, please, as a word of advice, demonstrate caution with color use.
We challenge you to work toward including only one accent color to your chart. And by all means avoid coloring each bar a different color! That’s a crime right up there with 3D effects…
If gray is just too much of a leap given the brand palette you need to comply with, then your ‘gray’ could be the less saturated version of your accent - just ensure there is enough contrast for it to be noticeable for the reader.
Hopefully this post has offered some tips for when you’re next looking to create a bar or column chart – or perhaps even when you are next reading one! This post was focused solely on bar and column charts, and is not cover stacked or diverging bar charts. We hope to work towards unpacking more about these formats in a separate post soon!
Look out for this and more posts with a similar angle in the future, all geared towards helping, inspiring and empowering designers to create clear and powerful charts and data stories. Our very own product, Datylon Graph, is soon to have the Beta launch – so if you are currently using Adobe Illustrator to create your charts, be sure to register and enjoy the new way to dataviz within Illustrator.
Datylon Graph is an extension that improves the charting power of Illustrator, so designers can keep working in their favorite tool. Along with keeping full creative freedom, Datylon Graph offers an continually enriched template library and safe and secure link with the data for painless updates when new data arrives. Check out full details here or sign up for the Beta today.
Until next time, happy vizzing!