Many organisations experience "Power BI sprawl," where a once-organised environment becomes cluttered with redundant reports and inaccurate measures. This article explains how to identify the signs of a messy environment and provides a strategic framework for restoring data governance and trust.
A positive way to look at this is, you’re not alone, and it means it was heavily used. Like a new kitchen, if it looks like it did when you first set it up, with everything in its exact place, pristine as the day you bought it, it probably means no-one's been using it. For most people, a fair number of the plates are chipped, half the plastic containers are missing their lids, the other plastic containers aren’t yours, the tongs are on their last legs, one drawer is overflowing with paper bags, and no-one is sure what these odd utensils are used for?
Power BI is a lot like that.
The lifecycle of Power BI clutter
Initially, you start with a clean Data Model including a handful of purposeful DAX Measures, a couple of Workspaces with a handful of Reports where each has a clear purpose and owner. Users are trained and encouraged to use the new Power BI environment with insights and functionality abound!
A few years in, and silos of information have appeared, the handful of reports have multiplied into over a hundred. People have concerns the data is inaccurate, or have used inaccurate figures in their board report. Others have stopped using it altogether, complaining it’s too hard to use, the handful of DAX Measures have exploded into a mess that people aren’t confident to use or can’t find the one they need to use for their report. The cycle from hundreds of Excel files, to handful of Power BI reports, has gone full circle back to hundreds of Excel files.
Why do Power BI environments become messy?
The flexibility of Power BI is both its strength and downfall. If people can create their own reports, they will, and over time you end up with 15 variations of the same original report because people don’t ask for the original to be modified. People create their own workspace for their team or department and a new silo appears. People create new measures without checking what exists or following convention so you end up with both a “Monthly sales by customer”, a “Customer actuals”, and a “Sales by period” which do exactly the same thing.
New datasets uploaded by users become out of date and incorrect figures are used. The proliferation of reports becomes a security risk, where new reports aren’t restricted to the correct users. People leave the organisation, leaving zombie workspaces and reports behind.
How do you fix a messy Power BI environment?
Like the kitchen, it requires some guidelines, some stewardship, and a bit of regular maintenance. Reminding people which drawer the tongs go in, updating the plastic containers, and a bit of re-training on those utensils people have forgotten what they’re for.
- Create a Data Dictionary: For Power BI, have a Data Dictionary that describes what fields and measures are for, which reduces new measures being created unnecessarily. It also captures knowledge so when people leave, that knowledge doesn’t go with them.
- Establish Self-Service vs. Core-Logic Guidelines: Have some simple Guidelines for Power BI usage, new reports should only be created for new audiences. New visuals or filters should be made to existing reports rather than creating new ones. Have guidelines on what changes to reports or measures that users can make ‘self-service’, and what requires approval ‘core-logic’.
- Conduct Regular Usage Audits: Set up a regular audit of usage, old reports can be archived if they’re not used after a period, rather than gathering dust in the corner of the kitchen cupboard.
- Implement Onboarding and Training: People cannot adhere to guidelines they don't know about. Ensure every new user understands where the "tongs" go and how to use the "utensils" properly.
Ready to clean up? Like a messy kitchen, all is not lost: pull everything out, take stock, work out what you want to keep, update, and throw out. Then sit back with a well-earned beer.
