Tim Messar is a Senior Data Architect working in the insurance industry at Sentry Insurance in Stevens Point, Wisconsin for over 12 years. He serves as a technical lead and resource on several projects and initiatives providing business intelligence solutions using SQL Server, Analysis Services Cubes, and Tableau. Tim is also an IT Consultant for Messar Consulting LLC where he enjoys solving problems and strengthening his skills regarding Data Science and Machine Learning. You can find him on LinkedIn or Twitter as @TimMessar.
Michael: Tim, you have been my saving grace in regard to dashboard actions and containers in the past. I’d like to follow a narrative where we focus on these two topics in our interview.
In your video book, MASTERING TABLEAU 10, you have two great chapters on actions and custom canvas control. What was your motivation behind writing these two chapters and why are they so important in Tableau?
Tim: Wow! Thanks Michael for the opportunity to sit down and chat with you. I have followed your datavizblog for years and I’m both honored and humbled to help with adding content to it. I have been passionate about data visualization for some time and it is great to see so many people contributing back to communities like Tableau.
Mastering Tableau 10 was my way of collecting much of what I have learned about Tableau in hopes of helping others. Much of data visualization to me is about providing the right focus and context at the right moment. At times, the “so what factor” of a Viz is present immediately, this is highly effective but can also be a limited. With static viz content, or an explanatory dashboard rather, audiences can easily understand what message or story the author is attempting to convey without interacting with the dashboard at all. Consider solely the images of Makeover Monday, without the need to investigate further, the stories are told as pictures, many as beautiful infographic scroll-style dashboards, assembled by authors from around the world.
However in a business setting, in my experiences at least, dashboards that provide up-to-date self-service analytics far outlast even the most amazing and gorgeous of static content. To be clear, there certainly isn’t a right or wrong way in my opinion, just different purpose I would say. In attempt to strike a balance, exploratory dashboards allow the audience opportunities to thoroughly move about the data, examine details, all in the flow and context. This was my intent for these two chapters. I wanted to present different strategies I have learned over the years that others can use to maximize the value and insight of their dashboards. I like when users spend hours in a dashboard engaging with the informative content. Actions and Interaction provide supplementary details and effectively more answers to more questions in real-time. Couple this with live data via shared data connections and visual best practices to communicate the content effectively and Tableau makes data come alive.
Michael: Can you tell me three of your favorite dashboard/workbook action Tableau Tips?
Tim: Sure, there are so many good tips/tricks with actions and many of them are so simple it makes me wonder why I didn’t think of it.
Some of my favorite tricks for actions include the following:
URL Action back to Tableau – so simple, but so powerful. Instead of loading up a workbook with dashboards and having tabs at the top to navigate perhaps, I create separate dashboards a link back and forth to them without the initial page load overhead. This is the same trick in dashboard we were discussing the other day. It can easily be used to create a PDF, CSV, etc.. in the hyperlink instead. Just like your use case of getting the data out of a tooltip and URL action combined with filter action can deliver. Isn’t it fun to experiment? I have used actions back and forth to link to Cubes or Google Maps or other more detailed web content. While I haven’t tried it (yet), I suppose one could even run a Lambda function via API Gateway from AWS which would open lots of new capabilities from a Tableau Dashboard URL action.
Publishing a dashboard with an action pre-selected is also interesting. This one is a bit sneaky but in the right scenario adds a lot of value such as a splash screen complete with directions on how to use to use the dashboard or other temporary focus on page load until the user clicks on the screen. I want to say that I learned this technique first from Mark Jackson in a fantastic set of Tableau Admin dashboards. In a specific one designed for Tableau Extract behavior, he provided a “current” look of the Tableau Extracts on page load, but once the screen was clicked the action opened up allowing the user to look back in history or search. It was a few cool and useful implementation that provided a helpful user experience.
Endless Navigation, one of my favorite action tricks, is an excellent way to move about the data as it deselects itself that Zen Master Joshua Milligan created in his post here. The concept is a simple one, but something that notably improves the user experience. His book, Learning Tableau 10, is an irreplaceable reference and “must have” for any Tableau or Data Visualization enthusiast.
Michael: Can you tell me three of your favorite custom canvas control Tableau Tips?
Tim: As much as I like to use actions to drill into more detail, filter, highlight specific data points of emphasis and other typical uses, it’s a challenge to push things outside of the box… sometimes literally. It’s fun to manipulate the dashboard in ways that almost make Tableau seem not like Tableau at all. While no doubt more effort to do so, creating button-like behavior generally works pretty well as viz-to-viz navigation where one viz is solely created not as a typical chart, but rather nav panel perhaps or better yet carry a color theme through the nav panel to the data points in lieu of a color legend (save that valuable canvas space).
Extending this thought a bit using containers allows users to have more granular controls and perhaps decide which vizzes are “anchored” in the containers and which move around. The end experience, as one example, could look like button click navigation with containers that aren’t bound to the visible dashboard screen at all. I have a dashboard that uses this concept sized something like 6000 pixels wide that slides into visual just the right sheets as the navigation chart selects them. I talk about this concept in Mastering Tableau 10, where we extend containers well outside the visible canvas space for added functionality. Your mileage may vary of course depending on the goal, but simply knowing the capability is available may be value add for the right dashboard.
Of course, regardless of the dashboard’s audience or purpose, having control over the canvas is something to strive for as a publisher. Just like the mindset of a front-end web developer perhaps, I wouldn’t want users to be able to do something that isn’t expected. Probably a more practical set of tips that I would highly recommend would include items keeping scroll bars off your dashboards, removing hierarchy behavior that could collapse or expand beyond control, or tooltip filtering that could leave inexperienced users looking for the Back button. It’s no doubt challenging at times especially with live, evolving data in a business setting to prevent squished or cut-off wording, evil scroll bars, ellipses (in headers, filters, parameters, legends), resolution handling, or other “misses” in dashboard design but these more simple tips will keep users focused on what they should see rather than overlooked blemishes in the design.
Over the years, I have come to very much prefer simplicity with the overall vast majority of dashboards I create. True at times it’s fun to overlay pictures or complex charting that more closely aligns with art than a functional dashboard, but there is truly something to be said about dashboards that have mastered the data-to-ink ratio with clear design, focused context, and effective visuals.
Michael: In working with your business partners at work, what seem to be their biggest pain points in using interactive dashboards in Tableau?
Tim: At Sentry Insurance, we have many talented Tableau users across the various departments and teams that recently our most notable “pain points” with Tableau was with our Tableau Server resources themselves. Several contributors in IT, including myself, offered solutions to better balance out resources, extract management, resolve poorly performing dashboards, etc.. So far the metrics are looking better and continue to trend in the right direction… and yes, of course I’m referring to Tableau on Tableau dashboards that we use within IT to monitor seemingly every aspect of our Tableau ecosystem from performance, admin tasks, operational metrics, data source management, permissions, usage patterns and even desktop license management).
Michael: What additional functionality can Tableau add, in the future, to Tableau Desktop to make the creation and use of dashboards better?
Tim: I would like to see Tableau continue to extend the “workbook formatting” capabilities, ideally to something that maintains my formatting style regardless of my workbook. Much like how custom colors and shapes are stored outside of workbooks in local folders, I would be interested in “saving off” my formatting style rather redoing it for new workbooks. As a work around because of this, I may copy/paste from a worksheet with my style to start my next dashboard only to drop it shortly after along with the data source it carried with… sort of like using a template or CSS in a way. This would be great for me at least; I tend to spend a lot of time fine tuning the formatting details.
Michael: What pain points do your users have prepping data to consume in Tableau and what tools, trick and techniques are you using to make this better for them?
Tim: In many cases, Tableau data originates from one of our ADS, Analytical Data Stores, at Sentry which is quite cleansed and ready for consumption. At times however or to quickly iterate/experiment, I’ve seen users blending data sources when they really could be using a cross database join to reduce the number of “data sources” tableau needs to load (unfortunately, this is still a sequential effort vs parallel processing of the visualizations added in 9.x). Outside of this, users may wish to look at the data differently where the structure of the data stores are only so helpful for example, when pivoting, unions or multiplication of the marks is required to achieve the desired effect. Maestro is sure to help with this of course as adoption of this new capability grows within Sentry.
To assist our Tableau publishers, Sentry Insurance uses a strategy of data source management that is quite effective on Tableau Server. It’s common to organize “galaxies” as we refer to them, spinning off the Business Objects “universe” semantic layer, with Tableau Server folders for easier permissions management, business friendly field names, metadata management such as embedded description for fields (or lineage or calculation logic if the logic was pushed down into the database or ETL). Similarly we aim to provide metadata about these shared data connections themselves in terms of content, time-slice of data, grain e.g., “what one row means” to clarify the data set and emphasize reuse among several dashboards.
Michael: So, the Tableau Conference 2018 in New Orleans is quickly approaching. What can Tableau do to make this an even better conference than before. What are some topics (sessions) that you feel have been missing in the past they should include this time?
Tim: I’ve been fortunate enough to attend several Tableau Conferences over the years. The conference is nothing short of amazing every single year. The sessions are well prepared, organized, and crazy insightful. It’s probably the nerd in me, but the atmosphere of the conference is something you truly have to experience for yourself, there is a constant energy and excitement in the air like no other. I’ll never forget my first conference in Washington D.C. back in 2013 (I think), I was completely blown away. Later at conferences in Seattle, I attended a session by Andy Kriebel who later invited me to a Facebook event where I also met several others in the Tableau Community like Joe Mako, Mark Jackson, and Anya A’Hearn all so skilled and talented. In Las Vegas, I met the incredible Joshua Milligan who helped provide technical review and Zen Master guidance for the Mastering Tableau 10 course.
It’s difficult to find much that I would change about the Tableau Conference. I feel there is something for everyone from beginners to the more advanced, business focused to the technical, managers to admins, and artists to analysts… the Tableau Conference will inspire well beyond the short days in New Orleans.
Michael: You have shown me a few tricks on debugging dashboard actions. Can you talk about a hypothetical checklist we should follow when we are testing dashboard actions?
Tim: Haha… great question! As you have watched me demonstrate, my approach to Actions is quite methodical. I like to add them one at time, name them clearly as to what they are, and step through testing one by one to ensure they work as expected. It would be great if Tableau would add a few more options to the Actions screen like coloring them or a place for comments perhaps or even a separate optional “debugging” window showing the history of which Actions fire with user interactions.
Another great and simple tip I use regularly is to use copy/paste when deciding what actions or functionality you like best. Just as Tableau allows copying fields, calculations, pills, data sources, and worksheets, there are times that I will copy the resulting Dashboard simply to play with Actions and fine tune it only to clean up later by deleting what isn’t needed.
Michael: Do you feel Tableau Maestro will be a better (or good) alternative to Alteryx?
Tim: I really can’t speak much to this question as I don’t have enough experience with either tool yet. I can tell you that my fellow Data Architect at Sentry Insurance who is also an Alteryx Ace in John Schneider would have his preference, but I do feel that Tableau thoroughly understands the need to integrate. I personally have enjoyed watching the enhancements over the years such as cross database joins, cross data source filters, unions (wildcard, etc..), and now most recently Maestro. Tableau’s attention to detail in this regard sets it well apart from other tools on the market.
Michael: Finally, what is next on your list of things to conquer?
Tim: I love this question, I have so many things that I’m completely fascinated by these days. Lately, I’ve been dipping my toes into several aspects of Data Science that I do not have much experience with using Python and Spark. Machine Learning, Deep Learning, and Computer Vision are a few topics that come to mind. There seems to be so many things to learn in this space and a tremendous amount of potential.
While I’m not sure I’ve “conquered” items on my list per se, I do feel I have had several great opportunities at Sentry Insurance that have allowed me to quickly gain experience, develop my skills, provide value, and even become a resource for others. Much like the theme of Mastering Tableau 10, learning is about the journey that doesn’t have an end.