Data governance seems to be the hottest topic at data-related conferences this year, and the question I get asked most often is, “where do we start?” Followed closely by how do we do it, what skills do we need, how do we convince the rest of the organisation to get on board and not see this as the team trying to “control” how they work?
The last question is a legacy of overzealous data governance workers from the 1980s, however the first few questions are an outcome of the data-driven world we now find ourselves in.
Moore’s Law has continued to hold true, and its exponential nature now sees organisations swimming not only in data, but new and external data sources. So where you start in 2016 is quite different from where people started back in the 80s.
For those starting out, trying to show value quickly can be a daunting task -- especially when the organisation sees data governance see as a necessary evil. So here are four steps to consider as you get started.
1. Don’t talk about data governance at all, if you can help it
While the definition of data governance is still correct, and the need for it is more important now than ever, I’d suggest you update the terminology you use with your organisation so that it's less about control and rigidity, and more about frameworks, guidelines and decision-making routes.
Many large organisations are still recovering from the hangover of allocating data governance roles back in the 80s, where they were executed with gusto and diligence.
Creating a new term for data governance is not easy, but using more friendly terminology is. For example, you could describe it along the lines of data policy management or co-ordination, and I’m sure there are a lot of other suggestions out there. The goal is to remove the idea of hard absolutes that used to come with this program and replace it with a different approach to managing data for the good of the business.
2. Focus on critical outcome-based activities
The second step is to focus on outcome-based activities. The need to govern data is based on the goal of ensuring we make the right decisions and take the right actions at the right time. Those decisions and actions should be based on the current objectives of the organisation.
Identifying the critical data underlying these objectives gives us our starting point. Obviously over time these objectives will change, as will the data we manage.
3. Set benchmarks and measure, measure, measure
Once these critical data sets have been identified, you must create data quality benchmarks. Creating a data quality benchmark is key in setting a stake in the ground upon which KPIs and measurement metrics can be based. It’s also the starting point in understanding which data cannot be relied upon in its current form. This is when data cleansing/standardisation needs to occur, to rectify the issues that currently reside in the data. This progress needs to be monitored and measured.
4. Set up the framework to manage data and exceptions
As a good colleague once said, the goal here is not to become a laundromat. In order to ensure data cleansing does not become a regular occurrence, guidelines and policies need to be put into place. These may be business rules for ensuring accuracy at data creation/capture, or it may be training programs for new users to understand why data needs to be captured in a certain way, and the fallout of not capturing it correctly.
Once these policies are in place, data governance frameworks can be updated to deal with exceptions. Exceptions are not necessarily wrong, hence the need for a framework to identify the cause and which level of the data organisation should deal with it.
When setting up a framework, use the above benchmark data to illustrate the “why.” This is key to the organisation understanding the need for what you're asking of them.
Over the years, I’ve found the benchmarking and measurement step by far the most valuable in terms of both providing a starting point from which to improve, but also in helping the organisation and the executives understand the fallout of unreliable data.
Providing evidence of problems using empirical data that’s attached to the organisations’ objects helps provide focus. It’s also key in identifying the ROI of a data project that requires a governance approach and allows a data team to quickly show value via the measurement of data on its journey to reliability.
However, this is only one side of the coin. Often those identified to become the stewards of the data are seen as the owners and problem solvers. And in an organisation full of people who are all attached to the data, that’s an unrealistic expectation.
For more on this, take a look at the following article: 5 models for data stewardship.