Project Governance of Self-Service BI, Without Killing the Self-Service

self-service BI


In utopia world, self-service BI leads to enthusiastic user involvement. Authors are decentralized in their areas where they know the subject matter better than anyone else, and work among the ultimate end users of the content they create. Dashboards are created and published in a timely manner, and meet the needs of the decentralized user community faster than anyone from IT/IS could ever dream to achieve. In many cases, authors are the same staff that are the scientists or other types of knowledge workers that now have the tools at their disposal to be able to build what they need and get it shared with others that need it. Let’s face it, with some of the reporting/dashboard tools out there (example Tableau 10), anyone with the ability to drag and drop widgets onto a canvas can build a beautiful dashboard in no time. In utopia world, not only are they empowered, but they no longer need IT/IS to slow them down. Utopia world seems pretty attractive. Right?



In dystopia world, what results is something like “the wild west”. It’s lawless and messy. Authors may or may not have the skills to understand how a data source works or how not to unintentionally misrepresent the data. Authors may not understand anything resembling best practice in terms of performance, creating undue burden on the server and data source. Authors create reports without regard to their prior existence, reinventing the wheel each time, creating lots of waste and duplicate reports to wade through on a very busy Tableau Server. Reports get published that have no discernable naming convention so that searching for a report becomes a less than optimal user experience. In dystopia world, IT/IS Server Administrators spend a lot of time trying to pin down which reports are causing the most drag on the Tableau Server and the database servers that are connected through the data sources.


The Sweet Spot

Self-service BI has many wonderful advantages (see Utopia). But if left unchecked and it is not managed, but rather seen as just another tool to roll out and offer up to users (ask my coworker about unmanaged SharePoint servers), it can lead to what will become an unmanageable world that cannot be lived in and that cannot be repaired. Our experience has shown that there is a sweet spot that will prevent Dystopia, and preserve the good parts of Utopia.


What we Helped the Client Implement to Prevent The Wild West

  • Governing Body Representing Decentralized Community

This group handles things like approving or modifying policy and procedure recommendations from us for defining who is allowed to publish, what process is followed to request that something gets published, what supporting documents that need to be created to have a report published.

  • Searchable Report Library and Report Request Intake

One of our deliverables was to build the client a fully searchable Report Library, that contains an intake mechanism for a report request if the search yields nothing. The intake collects information for a request and routes it to the developer staff’s TFS (Team Foundation Server) for tracking.

  • Publishing Procedure

Publishing is not a self-service BI function. It is performed as a request, only approved if certain guidelines are met (completion of associated documentation and peer reviews to name only two).

  • Training of Authors and Interactors

Includes content driven to introduce some best practices in order to prevent inefficient reports.

  • Maintaining and Distributing Effective Data Sources

Our experience has shown that authors that are not from the traditional world of relational or star schema report writers are not well versed at creating their own data sources, or even worse, blending them. In order to mitigate this, the vendor of the data mart built template reports that contain standard data sources.

  • On-Going User Community

Call it a user group, or perhaps even Tableau Tuesdays (yes, we will have tacos), there will be ongoing knowledge share and collaboration which we’ve currently scheduled twice per month. These sessions will also be good venues to distribute best practice information as it develops over time.​

Contact us to learn more about our approach to self-service BI and business intelligence in general: