You are preparing for a long road trip through the Himalayas. You have already outfitted yourself with all the recommended gear. Your vehicle has been serviced, and you are all set to go.
There is just one issue which you don’t yet realize: the only fuel that you have access to is contaminated.
Instead of enjoying the ride on a smoothly humming engine you will spend most of your vacation hanging around roadside garages, waiting for mechanics to try to figure out just why your jeep keeps sputtering up all the time.
Replace the jeep with your company and the contaminated gas with bad data and you can see where I am going with.
The implications of bad data and poor data governance
All enterprise IT systems like ERP, CRM, and supply chain management depend on high quality data to deliver results. E-commerce would collapse without quality product and customer data.
But companies have always struggled with keeping data quality consistently high.
According to Gartner Fortune 1000 companies lose more money because of operational inefficiency because of data quality then they spend on data warehousing and CRM.
In fact, the 1-10-100 rule from Total Quality Management can be applied fairly accurately to data quality, where it costs $1 to verify a record when it’s entered into the database, $10 to cleanse it, and $100 if no action is taken.
With self service BI and predictive analytics taking center stage in driving business decisions it’s even more urgent for companies to ensure that that raw data which these tools use is as clean as possible.
But the problems are not simply limited to bad data. Some other problems include:
- Absence of a data catalog which prevents end users like business analysts from quickly finding the data they need.
- Lack of universal data definitions and usage guidelines, leading to different teams interpreting the same data differently with confusing results.
- Inability of traditional data warehousing systems to deal with big data environments like Hadoop data lakes.
These are systemic problems that you will have to tackle before you can actually get tangible benefits out of your investments in BI and predictive analytics.
Getting started with data governance
An Aberdeen Research Group has pointed out that best in class companies were three times more likely to use data quality tools that other organizations.
To get a jump on the competition these companies use Master Data Management (MDM) solutions coupled with data governance processes. While the software itself is no big deal it’s the change in culture and organizational process that will have to be given higher priority.
The elements of data governance
Here are some recommendations to get started with data governance.
- Understanding data governance issues
Every company has unique needs when it comes to data governance. The first step to getting the process started is to understand where the issues related to data originate from. Once the source is identified a sensible roadmap can be drafted.
- Don’t leave it to IT
According to Forrester less than 15% organizations had business led data governance initiatives. This is a mistake because data governance is not tactical. It is a function that should be tied to strategic goals and initiatives. The top management needs to be on board because data governance is also about changing or tweaking operating models, processes and responsibilities.
- Integrate data sources
Data governance is hard enough with a centralized and structured data repository. The problem is compounded when unstructured data that’s diffused throughout the organization crops up. When Big Data enters the picture data governance HAS to be given top priority if the organization wants to deliver an excellent customer experience and thrive in the long run.
- Appoint data stewards
Many organizations have created the post of Chief Data Officer to own the data governance process and drive strategy around it. Reporting to the CIO, this role will become increasingly important in making sure that data is clean, consistent and accessible to end users.
- Build a common data vocabulary
Data inconsistencies can cripple timely decision making. As companies become data driven there is an urgent need to globalize the definition of metrics and standards within the organization. This effort can be led by the chief data officer so that all business units can implement change management process and transition to newer standards.
Implementing better data governance processes can radically improve operational efficiency, improve compliance with regulatory standards and wring out more insights from existing data. It can also make the organization more agile and nimble.
However, because the success is dependent on humans it will take dedicated resources and commitment throughout the organization for wins to be visible. There is a high amount of effort involved but the end rewards more than make up for it.