While the annual data-nerd conference (TC17) was underway, you may have seen me beaming from ear to ear as Tableau CEO Adam Selipsky busted many data myths during his keynote speech. One myth I was particularly pleased to see Adam bust is that, “Data governance means NO.”
Above: Tableau CEO Adam Selipsky speaking at TC17’s opening keynote.
As luck would have it, I spoke about self-service governance on a panel at TC17 with my fellow InterWorks colleagues Mat Hughes, Tony Kau and Derrick Austin. Between the four of us, we discussed performance, monitoring, governance and embedded analytics. Thank you to everyone who attended our session. Which leads me to today’s topic: how some of Tableau’s new features are enabling groups to implement self-service data governance with greater trust and ease.
The Wild West of Data Governance
At this point in time, if IT departments are still applying the rule that access to data is limited to a select few, those companies will find themselves struggling. They’ll realize they are not performing at levels they are capable of by leaning on their internal expertise and business acumen, which made them what they are today. Below is an abridged version of what I spoke about at the Tableau Conference.
Self-service analysis empowers the business users with the expertise to find insights in the data. They can free themselves from prescriptive reports that may only answer a small number of questions, requiring a protracted process when a new question presents itself. Self-service, however, does not mean a wild west of users connect to every table in a database, leading to the despair of database administrators everywhere.
Enter Published Data Sources
Curated data sources for trust and discoverability can now be produced by data stewards (those users who understand the database and business terminology). Often these two terminologies are not quite the same. Metadata is also important. Consider the following example. Below is an unclean data source (direct connection to database):
And here is an example of a clean data source (direct connection to database, cleaned before publishing):
So, what happened? In the clean data source:
- Fields are renamed to something business users are familiar with
- Formatted data types with % and £s for currencies
- Added common calculations i.e. profit ratio
We could also create folders if necessary and other formatting as required, but the above is only an example of what a curated data source would have applied to it. Tableau continues to embed through the enterprise fabric. What used to be done with consistent naming conventions is now available in 10.4 and it’s called data source certification.
Benefits of Data Source Certification
For an example of the benefits gained from this, let’s look at a Tableau-packaged workbook and its contents. Here are the elements found in a Tableau workbook. Your data steward likely knows where the data resides (its connection) and what the metadata should look like (date_1 is the transaction date). Data refers to the actual data being analysed, but the data steward may not be familiar with data visualisation techniques, best practices or understand the business well enough to know what questions to ask.
However, if we allow the data steward to create a certified data source, we allow our business user to focus on what really matters: the visualisation and finding insight within the data.
A few more advantages of creating certified curated data sources:
- Speed of analysis: If your user is focused on looking at the data rather than being worried about what tables to join, they will have more time for what really matters.
- Ease of use: Data sources business users are familiar with are key to increase adoption.
- QA only what you need: If data sources are certified, the need for QA is reduced. This applies a strict change-control process to the foundational elements. You can then be less strict with the visuals people build.
- Vetting increases with audience size: More people use the data sources and validate them, making it easier for others to use and rely on those data sources.
- Performant: Curated data sources have been tested and are less likely to cause performance headaches to the business users.
- Web Authoring: Fast parity with Tableau Desktop, allowing business users to analyse in the browser with limited loss of functionality.
- First point of contact: For many people, the first time they use Tableau is through the Server and Web Authoring provides a safe environment to learn in.
- Training: Tableau has released Web Authoring-only virtual trainings which are useful for those new to Tableau and self-service.
Final Word for Tableau Server Admins
Transparency is important if we’re going to successfully introduce self-service at an enterprise level. For those users who create content for others, either data sources or workbooks, it’s important to see how and when that content is being used, and if it’s still performant.
Admins should either grant access to a Tableau Postgres repository or create ways in which those metrics can be monitored. Not only does this provide a better synergy between IT and the business, it also provides accountability to those who create content. The users can then help IT by making sure the content they create is fresh, in use and performant.
Thanks for reading!