5 Eye-Tracking Discoveries for OPTIMAL Chart Design – Do you use these?

5 Eye-Tracking Discoveries for OPTIMAL Chart Design – Do you use these?


Bar charts have been around for many years. William Playfair who was a Scottish political
economist is considered by many sources to have created the first bar chart. In the last book he published, he also included
a column chart that looks like this. Since then, the visualization of these charts
have been improved on by the likes of Edward Tufte, who’s known for his concepts of maximizing
the data to ink ratio and phrases like chart junk and let the data speak. You can read all about this in his book, the
visual display of quantitative information. In this video I’m going to dig deeper into
column chart design and I’m going to share with you five tips when it comes to creating
the optimal column chart for any of your reports. Now what I’m going to be showing you here
is based on a study conducted at the university where I’m teaching. Link to the full report is below. The point of this study is to test if certain
design decisions that you make can influence the effectiveness of your graphs and here’s
the best part. They do this by applying eye tracking to make
an accurate assessment of what the observer is looking at. Here’s what it looks like. The tests are done in special settings. The way it works is you’re presented with
a question and you need to answer that question by looking at the visualization on the screen
while the camera is tracking your eye movements. Two parameters are measured. One is effectiveness, which means did you
get the answer right and the second one is efficiency, which means how fast were you
able to respond? Eye-tracking is supposed to provide more insight
into visual behavior. This is done with the use of scan paths, so
how you’re moving your eyes, your attention, on the screen and fixations what you’re concentrating
on. Here’s an example of an actual to budget comparison. Participants were shown this chart and they
were asked: what was the variance between actual and budget for the business segment
retail? This is an example of the resulting scan path. It shows a complex search pattern and lots
of eye movements back and forth. Another group was shown an improved chart
and was asked the same question, what was the variance between actual and budget for
the business segment retail? And here’s an example of the resulting scan
path. The eye has much less work to do, which not
only is more efficient, which means people were faster responding, but it also improved
the overall effectiveness, which means more people got the right answer. Now let’s get into the five tips. Number one, don’t use 3D. This is probably a no brainer, but I found
the results interesting. There were 84 participants divided into groups. They were asked two questions, does the actual
sales volume exceed the budget in June, and how high is the actual sales volume in June? Here’s the 3D version they were shown. For 2D, they were shown different versions
like these. For the first question, the study found, the
3 D graph had an 11% error rate while the 2D versions had zero. For the second question where they had to
guess the exact number, with the 3D version people estimated either way too high or way
too low, and if you’re still not convinced, I give you another reason to skip 3D graphs
and that’s the response times were 47% longer for 3d than 2d. Number two. Show the values directly above the columns. The study found that when they showed a graph,
with only labeled values, the results were better than using only a label axis. People were faster responding to questions
like how high is the actual sales volume in June? Number three, align label values vertically. If you don’t have enough space with large
numbers, we sometimes have the tendency to make them smaller to fit. Instead of putting them horizontally and making
them super small, we can put them vertically. Number four, don’t break the axis. This can result in lower effectiveness, which
means people misunderstand the message of the graph. Take a look at this and now this, do you spot
the difference? Number five, this wasn’t directly covered
in this study, but it’s super important. Don’t just give your charts a title, but make
them big and meaningful. This is the first place people look at before
they move their attention to your chart. They need to know what they’re looking for. I hope you find these tips helpful. Let me know in the comments below which ones
you’re currently using in your reports. If you liked this video, give it a like to
support my channel, and if you like to receive my weekly tips, push that subscribe button. Thank you for watching. See you in the next video. [inaudible].

100 thoughts on “5 Eye-Tracking Discoveries for OPTIMAL Chart Design – Do you use these?

  1. Very helpful. I normally truncate the numbers of they are too big (present in thousands or millions) instead of presenting them vertically I find it a bit odd reading numbers that way but it is an interesting tip.

  2. If someone want to confuse his boss then work alternatively. If further confusion is required then use secondary axis and remove the labels.
    If you want to show no differences then don't change origin, otherwise change.
    Good tips for dealing abusive bosses.

  3. Thank you so much Leila for your videos. I have improved a lot of my reports for the management thanks to your Excel tips. I will implement these 5 design tips in my reports to make them even more efficient and effective.

  4. My tip: Before I publish any report, I ask someone who knows nothing about the information being presented to give feedback. If it takes them longer than five mins to understand anything then I reevaluate my design

  5. Great info overall. If I may offer some feedback, your #4 tip is very vague, uses unusual wording (don’t break the axis?), and the 5 second comparison doesn’t highlight the difference. I think the takeaway is to scale the axis with the data to ensure the graph visually highlights the information and pertinent trends. Ie, doesn’t always have to start with 0.

  6. Obviously that was the most beneficial and helpful information in a video I have seen so far in 2019, or may be my entire life. As a data analyst in my department I am always seeking the best practice in data visualization in my report to help reader locate info easily and never get lost. thank you for this valuable video.

  7. @ Leila – What did you mean "don't break the axis?" Does this have anything to do with changing the default values?

  8. I find that the study referenced changes these tips from optional, personal preferences to proven techniques. I love the eye tracking charts! Thank you!

  9. Great video. BUT I am really surprised that is took this sophisticated study to conclude the obvious. HOWEVER, it only makes sense when you look at how many people create charts with these "mistakes". Not breaking the axis is extra important and many times the axis IS "broken" to intentionally mislead the viewer!!

  10. Hello Leila, great stuff as usual. On DONT BREAK THE AXIS, kindly expand. Are we you start from ZERO or another figure that then allows for a higher degree of accuracy if one had to guess the values (supposing they are no data labels).

  11. Very useful info indeed Leila. Up until now I'm always played around with charts to see what I think looks best…but now I can go straight to what actually works best. 👍

  12. Hello my very beautiful Leila, you are pretty every day that passes. I think it's philosophy, a new thing in your classes ….

  13. Thanks once again Leila, Nice to have study data to back up what we have been trying to impress on people for years. Especially like the NO 3D…nearly 50% slower comprehension. Probably even more for 3D Pie Charts. Be great if your Uni sends the results of the study to Microsoft for their Excel and Power BI development teams…

  14. Excellent video. The only bad thing about it is that it exposes the many mistakes I have been making over the years.
    I'm sure I'll review this video a few more times before building charts.
    Thanks!

  15. Oh Yesss….this is lovely! Thank you Mam. I appreciate you're taking care to include a bit of History as well.

  16. Hi Leila.. eye tracking.. that's cool. It's amazing how much chart junk is out there.. and from sources that should know better. Your 5 points are great and thanks to your various videos and courses.. are all familiar and in my tool bag. Thanks for sharing your amazing knowledge, skills and insights. Thumbs up!!

  17. Getting data is one thing, presenting it effectively is another. I like that you're talking about it more. Thanks for another great video.

  18. Don't use thick lines in black font for scale lines, make them very light, if required at all. Always try to use consistent colours for series between charts. Never use Excel's default colours for charts, they're horrendous! I use all the techniques mentioned, but interesting to know why I do though! Thanks

  19. Fascinating stuff. Trying to get my skill levels up in Power BI and some of the visualisations being bandied around the analytics world these days are just beyond baffling. I think all the analytics package providers would do well to subject their "fancy" outputs to your study – I bet I could guess the results?

  20. Very nice research in deed. I never use 3D charts for this only reason as they are not real 3D but psuedo 3D. For real 3D we need another axis data which is z (bubble chart and surface chart use this). More to know: https://en.wikipedia.org/wiki/Misleading_graph

  21. So there is also my professor that hates chart titles and data labels. She always tells me to delete chart titles and any data labels from my charts.

  22. Hi thank you for the video.
    I tried downloading the report. However I could find only the German version of it. Can u pls share the link of the English version of the report

  23. I wish there was a mandatory law to make those who create the awful presentations charts to watch this and learn the correct way it should be done. Thank you Leila.

  24. Thank you for sharing Leila, very informative and I'm using 4 of the tips, the last one is new for me (size of the title) but I will consider that in the future

  25. Some great tips! Like always. Thanks Leila. What I did missed and maybe was worth mentioning is a target value, this way anyone that see the chart can understand that the results are low or high.

  26. You thanked us for watching, we thank you for existing! (and providing us with all this)
    Thanks a lot for this, i can only imagine the number of UX pilot tests this took…

  27. Values above the bars makes it so much easier! Keep the great tips coming, and thanks for sharing your sources 🙌

  28. Love your videos as usual. Nevertheless, I noticed you spent a lot of time on the eye tracking issue and then cut down on the main points, particularly towards the end. Thank you though so much, as I find your videos to always be useful.

  29. Hey Leila, following on your video, I would change the title to something related to bar charts, instead of just stating “charts”. A little misleading I would say. Also, I would include markers in the video (like with music playlists, to find the next song) in case you want to just directly to the tips or to a particular one as a future reference.

  30. You're getting skinnier. Hope that's a good thing. I really love your tutorials. Very well researched and are very easy to follow. All the best!

  31. Great info as usual. I loved the idea of measuring the eye movement; it's so scientific. In general, 2D is better than 3D; but the 3D here wasn't of a great design.

  32. @4:27 I've seen that broken axis done to exaggerate the differences. I think it was a microprocessor company trying to make their product look way above the competitors but in fact the difference was irrelevant.

  33. Great advice! If only the UI team at Slack would heed this. My eye tracking there is like watching an earthquake at a ping pong ball factory. . . .

  34. Sorry, all are clear to me except #4. The second example always seems to be the better, but I struggle to read the vertical labels in 4. Do you suggest vertical or horizontal labels?

  35. Thank you for the tips Leila, I love your way of teaching, it just makes me naturally understand the concepts and techniques, and your online courses are just great, I am grateful

  36. I love this type of analysis which drills down to how people actually read and interpret graphs! Very insightful video Leila!
    Cheers, David from Australia! 😉👍

  37. Leila – thanks for the video – very interesting! Regarding tip #4, I may be misunderstanding your recommendation, but I generally do start a bar graph with 0 on the y axis, otherwise it can be misleading about the relative difference between the bars. Just my 2 cents!

  38. Dear Leila Thanks for all your teaching.
    Can we copy sparkline formula and paste in another cell as we do for other formulas ? PLEASE REPLY.

  39. The global eye tracking market was valued at USD 302.5 million and is anticipated to grow at a CAGR of 24.8% during the forecast period.

    Get more information feel free download [email protected] https://bit.ly/2ALNgG2

  40. Thank you Leila! For rule number 3, it is very important not to write numbers with a font size too small indeed but rather than write vertically, I prefer to format my numbers in thousands for example and always write them horizontally because it is our normal reading direction.

Leave a Reply

Your email address will not be published. Required fields are marked *