Tableau Insights via Larry Keller


Larry Keller, original manual writer for Tableau, now publishes more insights into the wonderful world of big data...

Part II - Jackson Pollock and Tableau - A Discussion

With a baseline established for our discussion, we now consider how Tableau is being used by the new authors and data artists.  The title or moniker data artist has become in vogue over recent times, we will focus on Tableau desktop users who author business related content (Worksheets and Dashboards) for content consumers. At the conclusion of Part I, I asked if this new wave of desktop users followed the old rules noted below. 

1. Do not be accidental - Think before dragging and dropping anything which was lacking in Pollock's "action painting" method.  What question(s) are you answering and for what kind of content consumer?

2.  Have a message -  Let the data do the talking.  The goal is not to become a "chartoonist" but to think visually - Tableau's Show Me options helps to render the message clearly.  There are others charts that the author can create. Do not waste your time trying to be cool.

3. Keep it simple - Brevity is rewarded.  Test your content with a colleague(s).  If they have to stare at your work like a Tableau/Pollock story you have failed.

4. Provide context - Well designed visualizations include and are not limited to correlation, trends over time and distribution. Do your data sources include critical information for a complete and compelling story? Example - perhaps population or data samples from economic sources is needed.

5. Use design fundamentals - Less is always best.  Do not abuse color.  Do not try to be cool....that is not the goal.  Like a good resume, white space is key to the creation of any visual that is to be understood in seconds. If you are building a worksheet that is destined for dashboard, think about the use of labels, filters of any type and the use of font size.

Now lets look at some examples:

1. Accidental? - Did the author intend to make the content consumer study the visual to understand it?

2. Have a message?  If after reading all the labels and associated marks, maybe there is a relationship between the price of gold and oil.

3. Keep it simple - No comment.

4. Provide Context - Yes, one can say that there is ample context. Time periods certainly.

5. Use design fundamentals - Color is abused. Assuming that this was in dashboard form, the content consumer would not need all the labels which are overwhelming.


  1.Accidental? I think not.The author (artist?)  spent considerable time creating what Tableau could render in Story Point but that option was not in the plan. 

2. Have a message? Mining gem stones at a costs to the labor forces. The visual should stand on its on without a huge font announcing 3.7MM lives.

3. Keep it simple - Delete the bar graph, isolate worst locations only, keep taking away and tell a simpler story.

4. Provide context - The author has provided context, the question is the value.

5. Use design fundamentals - Color abuse - the blue gradients get lost in a blue background.


  1. Accidental? No - Unlike Pollock the author almost used Tableau as if creating a PowerPoint slide without the benefit of a build process.  

2. Have a message?  Yes if you have a note pad to tie the pieces together.

3. Keep it simple -  Creating "simple" is a challenge. We could have had a meeting based on this visual.

4. Provide context - In this case, the amount of context is distracting.

5. Use design fundamentals - Each visual requires an annotation consuming space. The scatter plot is confusing at best. The map in the middle consumes space and results in clutter.

Comments are closed.