Visual Analytics: Keys to Success with Tableau
According to Wikipedia, Analytics is “the discovery and communication of meaningful patterns in data.” Although there are many Business Intelligence (BI) tools for analytics, Kenway has found that Tableau provides the means for easy data discovery, as well as a visual approach for communicating insights from your data. Analytics is important to business users because it enables reasoning and decision making based on data-driven evidence. Users can see and understand their data.
Tableau is a popular choice for BI and Data Visualization in the business community because it is exceptionally easy for non-technical users to explore their data. Tableau does not require business users to know Structured Query Language (SQL), a common programming language for querying data and managing databases, or data modeling techniques in order to explore their data. Instead, Tableau offers a rich user interface with drag-and-drop functionality that makes for easy charting and visualizations.
Here are my keys to success with Tableau to help you quickly discover and communicate your data:
Start with Questions. One useful way to start creating impactful visualizations is to ask a simple question and attempt to answer it using the tool. This ensures that you understand upfront what you are trying to communicate to your target audience. Let’s illustrate this with an example regarding the recent Major League Soccer (MLS) cup final (12/6/2015) between the Portland Timbers and Columbus Crew.
My question: Did the Portland Timbers and Columbus Crew load their rosters with expensive players and “pay” for a successful season to eventually get into the MLS Cup Final?
To answer this question, I gathered data from MLSsoccer.com on league compensation (re: expensive players) and league standings (re: successful season) and imported this data into Tableau. The following visualization compares both Portland and Columbus to the rest of the league’s teams by showing the relationship between total points (each win is 3 points; each tie is 1 point) and total compensation.
Here are my observations based upon this visualization:
- Like most of the league, both Portland (POR) and Columbus (CLB) had modest total compensation amounts (between $4 and $6 million). This means that neither team “paid” relatively more for expensive players.
- Teams such as LA, Toronto (TOR), and NYCFC have total compensations above $16 million (more than triple the payroll of Portland or Columbus) and, thus, have high payrolls with expensive players.
- By comparing Portland and Columbus to other teams, I can conclude that neither team loaded their rosters with expensive players to achieve their success (based on total points).
As you can see, I started with a question and attempted to create a simple, but impactful, visualization to help answer my question, and give the proper context (comparing all teams) to answer it. By starting with a question, I was able to begin my analysis with the end result in mind.
Performance matters. When adding new data sources to Tableau, users have the option to select between Live or Extract connections. A Live connection means that Tableau is connecting directly to the database or data source in real-time. An Extract connection allows for the use of Tableau’s fast, in-memory data storage. This in-memory storage is local to the user’s machine and is not constrained by the overhead of database connections or the slow performance of the underlying database. This is a key concept because although the Extract is not real-time, it allows for a great user experience when interacting and discovering data as you are not constantly waiting for queries back to a live database to load and can quickly see how visualization react to your exploration.
It’s a best practice to use the Extract connection when possible. If your business needs real-time data for analysis and decision-making, I would recommend investing in your database infrastructure resources, partitioning your data for optimized access, and performance tuning your database for best results. My favorite quote from Tableau’s whitepaper for Designing Efficient Workbooks is: “If it isn’t fast in the database, it won’t be fast in Tableau.”
Another feature that Tableau offers is the ability to go from an in-memory Extract connection, to a Live connection, and back to an Extract connection. This can help meet business needs should analytics on live data be necessary periodically. You aren’t stuck with the in-memory data engine architecture, and this gives added flexibility to your Tableau applications.
Here are some additional items to consider in regards to performance when using Extracts:
- Hide any unnecessary data fields to reduce the size of your Extract
- Consider using aggregation functions (e.g. sum, count, average) if your data is more detailed than required for your analysis
- Consider extracting a subset of the data (e.g. one month of data) to create your visualizations; you can always extract the full data set when needed
Quickly Explore Your Data with “Show Me.” Since Tableau offers a drag–and-drop interface for creating visualizations, users can easily begin to explore and display their data. Tableau’s usability goes a step further with the “Show Me” feature; this offers one-click visualizations based upon the data fields that are selected. “Show Me” highlights and suggests the visualizations that best match your data. Likewise, “Show Me” disables the visualizations that do not fit your data.
Let’s review “Show Me” with a simple example based upon the number of breweries in each state from the Beer Institute. Upon loading my data and selecting the data point for Brewers by State, the “Show Me” feature immediately recommends visualizations (those in color) and disables charts that don’t fit my data (those grayed out):
Based on these recommendations, I can quickly explore different visualizations and select the one that I want to use to communicate my data.
Here are three different visualizations using the “Show Me” recommendations:
“Show Me” allows one-click visualizations to see the data and I can ultimately decide the best way to communicate ‘Breweries by State.’
Effectively Communicate Your Data. A key to success with Tableau is to communicate your data well to other stakeholders and sponsors. In order to do that, you must start by understanding the objective of your analysis. I recommend that you focus on these simple fundamentals when creating your visualizations:
- Start with the “why”
- Who is your audience?
- What do you want them to know?
- What is your desired result (the “so what”)?
Once the objective is well understood and the right data is being used, the effectiveness of your visualizations will depend on selecting the right medium (i.e. visualization) and designing for aesthetics (making it clear and visually pleasing).
Key points to selecting the right medium:
- Keep your visualizations simple
- Use graphs when you are trying to compare entire sets of values
- Use tables when you want your audience to look up individual values
- Don’t go crazy with pie charts; bar charts often tell a better story about the data
Key points to consider for aesthetics:
- Don’t overload your visualizations with unnecessary formatting; keep your formatting simple
- Keep data labels simple so you don’t “take away” from the visualization
- Be sure to sort your data when using a chart
Lastly, effectively communicating your data is an iterative process and may require many enhancements and improvements. Tableau’s strength in quickly creating and analyzing visualizations really comes to the forefront in this aspect, allowing you to quickly create an initial product, present it to users, and tweak and refine it to get something that resonates. This will help you get to the end result: your target audience understanding the data and taking the desired action.
We love taking data and analytics at Kenway Consulting. If you would like to continue the conversation about Tableau for your business, or any other BI or Analytics platform, let’s get in touch: firstname.lastname@example.org.