Gartner estimates that 60 to 85 percent of companies’ big data projects fail. In many cases, poor data governance is to blame. Below, we’ll discuss some of the primary reasons why implementing good data governance can be so challenging.
Data governance projects at many organizations fail because the company is willing to pay lip service to the ideals of the project, but not fully commit itself to its success.
In many cases, data governance falls by the wayside as members of the governance committee get caught up in their day-to-day tasks. However, the business side has a valuable role to play in data governance: defining the rules, metrics, and KPIs that all employees must heed and monitor.
Other data governance projects fail because the committee is caught in one of two extremes: Either the program is so high-level that it doesn’t deal with the substantive issues, or it gets bogged down in trivial details that aren’t truly necessary, such as when committees use near draconian means believing they need “data cops” to control their “data citizens.”
When different parts of the business are receiving unrepresentative or unexpected data, it becomes much more difficult to implement a successful data governance program.
It’s all too common that employees in one part of the business will use fields and codes on a form in ways that the authors of the document failed to anticipate. This could be due to the fact that the form has not been updated to reflect current business realities, for example.
Whatever the reason, these gaps in expectations will create problems with data quality: how the data is actually being transformed and brought across the organization. With so many crossed wires and missed signals, it’s all too easy to throw up your hands in frustration at the difficulties in consolidating your disparate types of data.
If different parts of the business don’t share a common vocabulary, you may have inconsistencies with your data even when reporting the same KPI. Not only does this create a negative feedback loop of confusion and poor insights, it also makes it harder to launch a governance program in the first place.
For example, multiple users may define their own ways of calculating important business metrics such as profitability and revenue. Even worse, these calculations are being conducted in separate Excel spreadsheets that are disconnected from the organization’s centralized information assets.
Because these files aren’t in sync with the company’s standardized processes or its single version of the truth, different departments won’t be speaking the same language when they need to sit down at the table together.
Perhaps the greatest challenges to data governance programs are silos of data (i.e.: data silos): repositories of information that remain under the control of a single team or department and are not visible to the organization at large.
When certain factions keep their data isolated in silos from the rest of the company, employees are unable to benefit from the answers and insights that the data may contain. 80 percent of companies report that they have experienced moderate or high levels of difficulty with data silos.
Implemented correctly, data governance programs should help break down silos by giving the entire organization visibility into these fractured databases, and by defining processes to document how departments can share data with each other.
This blog post discusses the release of Oracle Analytics Cloud, a service that includes Essbase Cloud, Business Intelligence Cloud and Data Visualization Cloud.