"I despise grouped bar charts and recommend against them. "
Bill Shander
Author of "Stakeholder Whispering: Uncover What People Need Before Doing What They Ask", keynote speaker, workshop leader, LinkedIn Learning Instructor. Information design, data storytelling & visualization, creativity.
That was my comment on a great conversation-starting (and triggering!) post the other day from Enrico Bertini , and while I normally don't speak quite so plainly or strongly (well, sometimes I do!), I stand by it.
Grouped bar charts are never the best way to show any data for any purpose.
Yeah, pretty bold statement. Especially from someone who almost always says "it depends" to answer any question about data visualization. And from someone who believes there are literally almost no "rules" in data visualization.
But this one? Yeah, I stand by it. Let me say it louder. GROUPED BAR CHARTS SUCK!
Seriously, look at this grouped bar chart.
Now, tell me what you can learn from it. I'll wait.
Yeah, if you stare at it for a while, and look carefully at every bar, you may notice that unit sales went up (height of the bars) from 2022-2023 (from teal to pink) for the first two sets of bars, but went down for every set after that. But it isn't intuitive, is it?
Now, try this visual.
领英推荐
Isn't it now IMMEDIATELY clear that sales went up for those first two and went down for the others? Sure, in addition to changing the chart type, I also changed how color is used (to show growth vs. shrinkage instead of one year vs. another). This also really helps.
But, I'll say it again, I don't think I've ever seen a grouped bar chart that couldn't have been replaced by another chart with a better outcome.
So, why do we use them? Because our stupid tools make it too easy to do so. And because we're too lazy to really ask ourselves if each visual we create is the BEST way to show what we're trying to show. In fact, do we even know what we're trying to accomplish with every chart we create?
No...and that's actually the subject of my book (coming in April), but that's a conversation for another day!
For now, I ask you...I beg of you...the next time you make a grouped bar chart, just pause. Look at it. Show it to a friend. And ask yourself and them if it really is the best way to show what needs to be shown at that moment. The answer, I can tell you already, is "no", so try something else!!
(BTW, to be clear, I have zero scientific, research-backed information to back up my claim. This is my opinion. In fact, that was Enrico's primary question for his post - asking if there is any real research about the effectiveness of grouped bars. I hope someone takes up the task. And hopefully proves me right so we can abolish them forever!)
==========
Pre-order my book, Stakeholder Whispering, which comes out in April!
Ask me about virtual or in-person data storytelling and visualization training for your team
President at Peltier Technical Services Inc
3 周You found a poor chart choice, which the "designer" (was it even designed) made worse with their color choices and data labels.
Instructor and best-selling author, data visualization and dashboard design | Taught in 15+ countries | Lecturer @ Yale, Columbia | LinkedIn Top Data Visualization Voice
3 周Bill Shander Steve Wexler Enrico Bertini I agree that clustered/grouped bars are virtually never the best choice when showing *time series* data, as in your example. I think they can, however, be the best choice when showing *non-time-series* data, specifically, when it's important to precisely compare the parts among several different totals. Or would you disagree?
Night Auditor @ Alpenhotel Montafon
3 周oh my god! the humanity! who, what, why, huh? ?? NEVER HAVE I EVER hated charts more than these ones! absolute nonsense, the hardest to read and make sense of, and one of the very reason I won't get into an article is when I see these charts rear their ugly head towards me! yuck! there must be a government mandate to abolish these vile things ASAP, and if there is one thing left for Trump to remain as one of the forefathers of modern America, then THIS should be his stepping stone! and I am NOT kidding!
Founder | ROI Strategist | Marketing Analyst
3 周Good insight
Independent Data Visualization Consultant | Microsoft Data Platform MVP
3 周Oh goodness. I vehemently disagree on this one. If a chart sucks, it's usually because it's not the most appropriate for the data/insight. The chart doesn't suit the data in both instances. Both charts are fine (except perhaps the colours and labelling choices), neither are the best choices for the data and narrative.