I've become quite the Tableau enthusiast in recent years. Since my time at U.S. Soccer, I've been using the business intelligence tool more and more regularly, transforming internal reporting from clunky and tedious to beautiful and efficient. In my current role as a consultant supporting Naval Special Warfare's Human Performance Program, I use Tableau to visualize the program's efficacy through overall program utilization, injury and performance tracking, and return on investment, amongst other things. In total, I've developed over a dozen interactive dashboards that answer a variety of questions, provide insights, and enable immediate and actionable decision-making power for our program stakeholders. This past year, I even attended Tableau Conference 2018 (which had a staggering 20k+ attendees), and, in addition to getting an opportunity to party in NOLA's Mercedes-Benz Superdome, I earned my Desktop Certified Associate certification in the software.
First of all, if you're unfamiliar with the power of Tableau, check this out to start. This article highlights the three finalists in Tableau's yearly competition, Iron Viz, with global weather being the overarching data set. These three, dare I say legends, successfully built their dazzling visualizations in a mere 20 minutes (they prepped for weeks/months). Sports fan? This one visualizes Clemson's 35-31 win over Alabama in the 2017 CFB National Championship. Or if you're doing a deep-dive analysis on, say, car accident data in NYC, this would be a good place to start. All in all, what I'm saying is this: Tableau is freakin' awesome, and its capabilities know no bounds. If you want to bring data to life and give your audience a chance to interact like never before, then Tableau is the way to do it.
If you're reading this and teetering between whether an investment in Tableau is worth it, my opinion (totally unbiased, of course) is this: ABSOLUTELY. I've also used Qlik Sense and Power BI, and while each has its positive attributes, Tableau is by and far the industry leader in business intelligence software. A few traits that distinguish Tableau from its competitors are: ease of use, performance efficiency, beautiful visualizations (this matters!), the online community, mobile compatibility, and complex, out-of-the-box solutions *. Perhaps I'll write another blog post comparing BI tools in the future, but I'll move on for now -- but feel free to reach out if you have any questions.
Now, to the purpose of this blog: to discuss a few tips and tricks I've picked up in my use of Tableau that I believe are worth sharing. Several are somewhat simple and maybe even obvious, and a few others have niche use cases. In any sense, take what you need and leave the rest. Enjoy!
1. Use KPIs/BANs
One of the most effective tools in building a valuable dashboard is the use of KPIs (key performance indicators), also referred to as BANs (big ass numbers) in the Tableau community. When analyzing how a reader perceives a dashboard and where his/her attention first goes, it's generally either to the upper-left corner of a page or to the KPIs that illustrate the importance of the data we're visualizing. The remainder of the dashboard should effectively explain what those metrics mean and why they are important. I highly recommend Adam McCann's KPI visualization post as a good starting point in incorporating them into your dashboards. Ultimately, using BANs will give your viewer immediate, quick insights into the data without a need to explore too extensively. Above are two excellent examples from McCann's post showing how BANs can be used to convey current metrics and performance over time.
2. Blend > Join
Okay, this certainly isn't true in every case, but I've found data blending to be an absolutely vital tool that no other business intelligence tool offers. In short, Tableau enables you to create data relationships based on primary keys/unique identifiers (e.g. ID, SSN, etc.) without the need to join data tables together. This can be valuable for two reasons: 1) the back-end processing speed is much more efficient since the blending is completed before any tables are "joined", and 2) not all data sources allow for easy joins based on table structures -- aka less pain for you. For example, in my dashboards, I blend Navy operator names and DoD IDs across dozens of data sources to essentially view a comprehensive landscape of their performance in one dashboard: program utilization, injuries, assessments, and so much more. With this, Tableau automatically recognizes the fields you've identified as having a relationship and does the heavy lifting for you. For a better understanding of blends vs. joins, check out the article here that does a much better job of explaining it than I ever could.
3. Viz in Tooltip
In too many cases, developers want to fit too much into one dashboard. This dilutes the overall intent of what the dashboard is trying to communicate and can leave a viewer confused or overwhelmed. To mitigate this issue, Tableau has developed Viz in Tooltip, a feature that enables you to embed a visualization/chart directly into the tooltip (view when you hover over a data point). This is an amazing, yet simple, tool that allows for a deeper dive on a data point without effecting the primary visual. As an example, you may have a geographical map demonstrating the overall profit of each US state using a diverging color scheme, and when you hover over each state, the view may populate with that specific state's profit over time in a line chart. A quick tutorial can be accessed here. In truth, this is one of the easiest solutions that Tableau has to offer, but it's one that is more about quality and necessity. If you decide to utilize this feature, do so sparingly, namely when you are limited in dashboard space and can dive deeper with additional insights.
4. Collapsible Viz
The purpose of this trick is very similar to use Viz in Tooltip: to add additional drill-down details without overwhelming the viewer with too many sheets on a dashboard. However, whereas Viz in Tooltip only appears on hovering over specific data points, Collapsible Viz (as I call it) appear and disappear on the dashboard when selecting (and deselecting) a data point. A good use case here would be in real estate. Say you have a regional map showing the average listing price of homes in the area; upon clicking on a specific neighborhood in the map, a detailed table of all of the homes in that area might appear below the map -- magically out of thin air. Deselect the data point on the map, and the data table disappears from sight. Having the ability to hide and display different sheets with a single click creates a powerful effect and enables the reader to gather additional insights without a cluttered view. To be honest, implementing Collapsible Viz took me some time as the process involves understanding how to organize layout containers and dashboard actions -- but it's a worthy investment to learn and incorporate into your dashboards while saving precious real estate. For a quick tutorial, read Dustin Wyer's post here.
5. Custom Image XY Mapping
This might be one of the most powerful visualization tools you can incorporate into a dashboard -- if you have the right use case. Whereas Tableau has already mapped geographical data based on city, state, zip code, and so forth, there are dozens of other ways you can create custom XY coordinates to map data points to specific locations on a visualization. In my role, we want to analyze injuries by body region; instead of a simple bar chart showing each body region, why not map the frequency and severity of those injuries ON an image of the human body? Or let's say you wanted to show the most frequently replaced parts of a vehicle; wouldn't it be awesome to show that data on the vehicle itself? In the example and tutorial by Tableau legend Ryan Sleeper, he maps data points to show defensive player locations on a baseball field. This process is actually relatively simple and only requires a few things: finding the right use case (and then stock image), custom mapping XY coordinates for each data point, and updating the axes to fit your data. From there, it's simply a matter of adding your measures/dimensions and customizing your visualization. Above is my example using Navy injuries (using dummy data) mapped against a stock image of the anterior and posterior human anatomy.
6. Action Drill-Downs & Buttons
Within Tableau, dashboard actions are the money makers. They allow for hover/select filters and bring the data interactivity to life. When you select a data point on sheet one, the rest of the dashboard updates to reflect the newly selected data set. Actions have a lot of added value, but two fun use cases for action buttons are 1) new dashboard drill-down and 2) buttons. For the former, it's often valuable to have executive-level dashboards that show the 10,000 foot view; key metrics, trends, and overall performance. However, adding actions to these sheets can reap immense benefits if there's a need for an executive to drill deeper into the data. When adding your dashboard action, simply change the target sheet(s) to those on a different dashboard, one that provides more context to the original viz. A simple video tutorial of this feature using tooltips can be found here. The second part of this is a bit more complex but can provide substantially more value as you can begin adding URLs and other unique dimensions. In my Human Performance dashboard, for example, I've added buttons on a home page that link to each of the workbook's dashboards (e.g. Utilization, Injury, Operator Report Card), and each of these subsequent dashboards has a "home" button (which is fittingly the Naval Special Warfare logo, maintaining a uniform style) that returns the user to the landing page. It's a simple trick that adds style and functionality. To learn how to add buttons, review Ryan Sleeper's quick guide here.
7. Adopt Best Practices
The last tip, or rather advice, I have is a very simple one: ADOPT BEST PRACTICES! I constantly find myself reviewing Tableau Public to find inspiration for new visualization ideas. What are others doing to bring their data to life? I'll often open up a new workbook on a completely unrelated data set and identify several key takeaways that I can incorporate into my own work. Sometimes it's stylistic, other times there's a cool trick that will wow my end user, and almost every time I find another way to optimize my dashboard's processing speed. Additionally, I'm searching the web for user guides, tutorials, videos, and forums at least five times a week to improve my own work and learn new skills. Like with programming, the beauty of Tableau is that there is a massive community of users who are experiencing similar problems, and that same community is exponentially posting new content and solving users' greatest challenges. If you're new to Tableau and experiencing difficulties, check out the Tableau Community Forum, and if you want more inspiration on unique ways to blend data and art, I'd highly recommend the Tableau Public Gallery. There's ALWAYS more to learn. Happy vizzing!
BONUS: Sankey Diagrams
To be totally honest, I was done writing after my seventh tip, but I wanted to share the COOLEST chart type imaginable: Sankey Diagrams. These should be used exclusively to demonstrate the flow of data. In the most common use cases, Sankeys can be used to show things such as energy flow or college degrees to different careers. However, there are some other unique examples. Highly regarded as one of the best data visualizations ever, this Sankey diagram showing Napoleon's forces over time going to and from Russia. In my specific use case supporting NSW, we track referrals to and from different components of our program (e.g. Performance Nutrition, Sports Medicine, Behavioral Health) to identify the holistic, integrated approach of the program (example shown above, using dummy data). This is a complex chart type and will certainly require some practice, but if you're eager and want to take it on, follow along to this user-created tutorial video -- you'll be happy you did.
Opmerkingen