BI Tools

Tips and tricks for building information maps, OLAP cubes, reports, and dashboards

BI Admin

Learn your way around a SAS BI installation.

Visual Analytics

Learn your way around the SAS Visual Analytics tool

Coding & Data

Extract, transform, and load your data into the SAS BI toolset

Stored Processes

Create and design stored processes like a rock star

Home » Data Visualization, Visual Analytics

Data Visualization: How to Avoid an Epic Fail

Submitted by on 2013-10-10 – 10:47 AM 6 Comments

As I prepare for my upcoming A2 Academy Visualizing Your Big-Data: What to Use, When & Why lecture, I was looking for some examples of effective and ineffective data visualizations. The lecture is part of the All Analytics Big Data series, which is online and has no registration fee.

A few posts ago I shared an example from a humble marketing dashboard and today I’m sharing a graph from a trouble ticket system. To its defense, the author was showing how easy it was present grouped information with some code.  I’m not sure if the author was seriously trying to communicate anything more than that.  Just so you know I am not always picking on sad little charts.  🙂

Answering Questions with Data Visualizations

Effective data visualizations have some common characteristics, which is communicate a message clearly and quickly. It is always fun to find one that makes you scratch your head.  Usually when you find an ineffective data visualization there is a marketing message [look how cool this is! or even you can do this!] not far behind.

Authors use data visualizations to answer an implied question or at least clarify a point.  For instance, a manager might want to convince his boss that more support staff is needed and prepare a line chart showing the ticket arrival trend. In essence, the chart answers the question “Are more trouble tickets being opened?” or “What is the trouble ticket trend?”  or even “How many trouble tickets were opened last week?”

What is the Question

I want you to study the following data visualization for a moment and ask yourself, “What question does this data viz answer?” or “What point does it clarify?”

Click to see larger image

Click to see larger image

Was it “What is the ticket arrival rate?”

At first I thought the question was “What is the ticket arrival rate” or “Which day is the busiest?”  These would be logical questions for a manager trying to plan the staffing hours for a Support department.  The chart appears to be time-series related with weekday as the time and sort of indicated by the title Number of ticket raised in the last week, per day by department.

So a quick analysis would be “The first of the week is rough, then it heats up again at the end of the week.”  However, if you examine the x-axis you might notice that the days are not in sequential order (Sunday, Monday, Tuesday, etc). Then I thought, “Oh they must be ranking the days by arrival”.  However, Friday is listed first and it was not the day with the most traffic.  If the days were ranked then Monday would be the first day.  I cannot really tell from this chart what the second biggest day is without more study.

Was it “Who opens the most tickets?”

Maybe the chart really meant to emphasize the department, perhaps the question was “Who opened the most tickets last week?” This would help the manager understand if one department needed additional staffing.  It’s fairly easy to see that Support (blue line) opened the most tickets, but which department was second?  Really, if you had to rank them, in what order do they belong?

Just like with figuring out the ranking of the days, you have to start counting in your head. If you are the manager reviewing this chart – it’s costing you time. How would you grade this chart on “efficient communication” when the manager needs a calculator or notepad to figure out what happened?  [More about elements of effective data viz here.]

Have you determined what is wrong with the axis yet? The creator used the defaults and the software placed weekdays in alphabetical instead of sequential order.  This is an example of what can happen when you let the software do the work for you.

Data Visualization Makeover

Is it possible that this humble data visualization is trying to communicate too much information at once?  And ends up doing nothing but creating confusion. There are 2 questions: What is the daily arrival rate and Which department opened the most tickets? I think it’s confusing to have it in one chart – you spend too much time trying to figure out the information.

Here’s my suggested makeover using SAS Visual Analytics and some interactive reporting. Let’s put the data in one panel but use separate charts.  In this example, the top chart shows the overall arrival rate for the week and the bottom charts provide details on who opened the tickets and the main issues.  If the manager wants to see the week before, she can select it from the drop-down.

Also I added a reference line that contains the average tickets opened per week as an example.  You need a baseline to know if you are getting more or less tickets than you normally receive.  Probably a line that showed the average per day would be more useful since there is variation.

sas visual analytics interactive reporting

Click to see larger image

What I like about SAS  Visual Analytics is adding the interactivity.  When you look at this data, it’s noticeable that Monday had the highest tickets.  The  manager can click on the point and the report instantly filters to show the department with the most tickets that day and their associated issues.  Support opened the most tickets and in general MS Office was the main issue.

sas visual analytics interactive reporting

Click to see larger image

For more details, the manager can click the department name to determine the issues.  By hovering slightly, you can see there were 92 issues opened by Billing that day.  And when you click the bar, the issues chart updates to show the main issues.  Nice.

sas visual analytics interactive reporting

Click to see larger image

I think I would probably add a link to the Issues that allowed the manager to see a list of the open ticket or maybe their current status.  Also I would want a tab that showed the fix response time, overdue tickets, and other similar metrics.


Even without the interactivity, when I split the chart into two charts it made it easier to understand the information at a glance.  The interactivity really made the data easier to understand and answer questions on the spot. Plus … the  manager could use a tablet to review this information while waiting for a meeting to start.

What do you think?  What might you have added or taken away?

Never miss a BI Notes post!

Click here for free subscription. Once you subscribe you'll be asked to confirm your subscription through your email account. You email address is kept private and you can unsubscribe anytime.
The following two tabs change content below.

Tricia Aanderud

Director of Data Visualization at Zencos Consulting
Tricia Aanderud is a SAS Business Intelligence and Visual Analytics consultant based in Raleigh, NC who works for Zencos Consulting. She has written several books about SAS, presented papers at many SAS conferences, and has been using SAS since 2001. Contact her for assistance with your next project.
Spread the love

Tags: , , , ,


  • Tricia says:

    That would be awesome. After I read this comment I looked at other visualizations and thought the same thing! (You’ve tainted me Bob!!) 🙂

    We need to send that suggestion over to SAS Product Mgmt.


  • Bob Whitehead says:

    Hi Tricia,

    I really enjoy these walkthroughs on improving visualisations.

    My only “addition” I would like to see (if possible) is that when clicking on any data point to subset the data, that the title or a subtitle in the “drill down” view should reflect information on the drill.

    eg: The Department chart should show which day I am seeing in the Department data and in the Issues chart should show the Weekday and Department I am viewing Issues for.

    This may already be an option on linked drill downs.


  • Quentin says:

    Great post! And I totally agree about the box plots. For the past year or so, if someone asks for a plot showing means, I usually give them box plots on the first pass. It’s amazing how much more information they pack in. And it’s rare that people don’t like them. The ineractive nature of this stuff for drill-downs etc is just amazing. Can’t wait for your book!

  • Analytics Maturity – definitely! I think this has evolved over the years, through education and software that can easily generate box plots. I also believe box plots will be the mainstream visual (instead of bar charts) before too long. Especially as school students are being taught how to interpret them!

  • Tricia says:

    I agree about the box plots. You make an important point … audience level is important. Sometimes I wonder as data analysts if we discount the users slightly. I didn’t start out with statistics background but the first time I saw some of our customer service data in a box plot … I was astounded. The box plot perfectly expressed what was happening with our tickets.

    After our local statistician showed me, I was able to show the other managers and we all got a Stats Boost! As a management team, we learned that infrequently we had tickets that went outside of the aging window. We were able to tighten up our processes in the right places because we didn’t have to change for the 1% of tickets that took longer than 60 days to fix.

    It’s part of how the organization moves forward on I guess what I’d call the Analytics Maturity process.

    What are your thoughts?

  • Hi Tricia,

    Great walkthrough on how to improve your data visualization. Something to keep at the back of your mind, which you’ve shown, is to question the visual… If you were a stranger to the data, what is it telling you? Can you easily interpret it?

    Loved the interactive report you generated. Personally I’d also like to gain further insight via a box plot to examine the distribution of the tickets opened by department. This way a further comparative analysis between the departments can be made with percentiles and averages. However this may type of visualization may be more suited to an analyst/data scientist rather than a manager/executive and as you point out, further metrics would be useful in another tab (section) of the report.

    Not only does one need to choose an appropriate visualization but also to think about the consumer of the visualization and their skills and ability to interpret it.

    Looking forward to your A2 Academy lecture and SAS Visual Analytics book!