This blog post originally appears on the BI & Analytics Conference website and is reproduced here with their permission. Would love to see you all there if you’re available.
With the mass adoption of departmental visualization tools (think SAP Lumira, Tableau, Qlikview, Power BI, etc.), the spotlight on data governance is really heating up as organizations race to make sure the data finding its way into those tools is clean, consistent, and correct. While explaining data governance is easy, putting it into place often isn’t.
Data governance is a defined process an organization follows to ensure high quality data exists throughout the complete lifecycle. The key focus areas of data governance include availability, usability, integrity and security.
So, why do we need data governance? Because we can’t be an effective data-driven organization without it. Missing any of the key focus areas highlighted above can seriously expose your organization.
- Availability – I don’t really feel I should have to drive the importance of data availability home with this particular audience. If it isn’t important, we’re all out of work.
- Usability –With the increased use of data in people’s personal lives, the bar for satisfying business needs is only getting higher. I hate to have to say it, but people want to interact with their work data the same way they interact with intuitive social media sites (I am not looking at you, Snapchat).
- Integrity – Just imagine telling 100 people they’ve lost their jobs due to missing budget, then realizing that it was just someone in accounting fat-fingering the electric bill.
- Security – There is literally no end to the data that you can accidentally expose (HIPAA, PCI, payroll, etc.) which can get you fired and/or thrown in jail. Be careful out there, folks.
So now that we know why we need data governance, what does it actually look like?
- We need a single version of the truth. While vendors have been marketing to this since the 20th Century, it is still a huge pain point for most organizations. Different analyses will come to different conclusions, but we need to at least try to make sure the data each analysis starts with is correct.
- Business owners must own the data steward role. IT needs to properly enable them, but we’re not going to definitely know that a particular expense needs to be mapped to a certain account. That’s a business decision and they need to be held responsible for that.
- Our governance must be transparent. I had a real light bulb moment a few years ago when I realized just how integral two types of transparency are to a governance process. First, the metrics but be thoroughly documented and consistent across the enterprise – tribal knowledge isn’t enough. Second, you should be able to drill from the high level metric all the way down into the details of a number and each of its components to get real buy-in on their accuracy.
- It’s about the journey. While the big current push is to use machine learning to analyze your data without any human hands, the truth is that there is real value to your organization to go through the descriptive > predictive > prescriptive lifecycle on each new metric. While the thought of reading the last page of the book is tempting, you miss out on a lot of trust, camaraderie, and truly understanding your business by going through the process.
The one last thing that I want to point out is that data governance is about the people – strong analysts who understand how to use it, strong leaders who will push back on being handed a spreadsheet that disagrees with their enterprise dashboard, and strong data stewards to make it all come together – more than the technology. Fortunately, you can hear about both at the 2018 BI+Analytics Conference.