Three steps to ensure successful data governance

Featured article by Ulrik Pederson, CTO of TARGIT

In many companies, the data governance strategy is one of an IT gatekeeper. This traditional approach—centralized, strict, and secure—is still valid for enterprise multidimensional data warehouses. But it isn’t right for every organization. To promote innovation and experimentation among teams, a new way of handling data governance is needed. Here are the three key steps businesses should take to ensure data governance helps rather than harms their overall BI strategy.

1. Ensure data quality

Before a business begins implementing new strategies, it first must ensure the data it plans to work with – both internal and external – is accurate. An analysis is only as good as the data behind it. When implementing a BI strategy, data must be accurate, authentic and trustworthy. Customer data, for example, doesn’t work if there are different versions of that customer within an ERP system.

Data cleansing can be quite a lengthy process, though. IT should gradually increase the quality of data along the way as data proves to be useful to the organization, ensuring short-term ROI compared to what would typically be a massive data cleansing project of new data. This process is called “sandbox analytics.” In other words, break up small, isolated groups to produce, experiment with, and share data before considering wider adoption.

While administrators should have access and control over applications, data governance is about more than just permissions and settings. As an increasing amount of unstructured and semi-structured data from outside traditional data warehouses is pulled in, data governance must include a strategic, trustworthy method for dealing with data quality. For many, this means considering a bimodal BI strategy.

2. Incorporate traditional and modern BI strategies

An analytics solution must accommodate agile, self-service experimentation and data discovery as well as the traditional reliability and security of data; this is known as a bimodal BI strategy. A bimodal BI strategy should not only facilitate traditional business operations—the classic data warehouse and continuous decision loops—but also discovery and innovation.

In order to be successful with external data analytics, experimentation is a must. The ability to play with big data sets and analyze them on top of what is already in the data warehouse encourages employees to think strategically without the need to pull in IT.

3. Embrace the centralized and de-centralized

Once an organization adopts a bi-modal BI strategy, it can eliminate traditional business intelligence disciplines seen through the lens of “either/or.” Bimodal BI is both centralized (company-wide initiatives) and decentralized (agility, innovation, and exploration). The benefit of this approach is if users with domain knowledge create new data models that prove valuable and should be made accessible to the organization, IT or the Business Intelligence Competency Center (BICC) will have the option of promoting the new content so it can be used across the organization.

With bi-modal BI, administrators can adopt a strong data governance strategy by leveraging a BI solution that permits control over the server and applications for use across the organization from a single control panel. This user set should also have the ability to identify permissions and control who has access to what data for everyday analytics use.

A de-centralized, relaxed, creative data governance strategy is necessary for any type of ad-hoc data discovery. And, in order to become a data-driven organization that makes decisions based on facts, not gut feelings, companies need to set data free to employees in every role throughout the business. A robust data governance strategy is the only way to do so properly and securely. Without it, companies run the risk of making decisions based on data that is incomplete or incompatible. Data governance is the assurance that every decision is based on the right valuable data – that is what it takes to be truly data-driven.