GRAPEVINE, Texas—Master data management is all about control and governance, making sure that there is one set of truth that helps feed clean data to important business processes.
That’s why big data scares the administrators tasked with keeping master data clean: the variety, velocity, volume and complexity—to use Gartner’s definition—make it almost impossible to control.
But big data can actually serve as a complement to master data management, changing the way executives understand data’s role in the enterprise and helping the master data management system adapt more rapidly to new trends.
Just know that the emerging sources, properties and behaviors of data are unruly, and there is nothing you can do about it.
Master data is any non-transactional data that is critical to the operation of a business—such as customer or supplier data. Master data management is the process of managing that data to ensure consistency, quality and availability. New sources of data, such as machine sensor outputs or entries on social networks, are challenging this field.
“Big data is an invader,” said analyst Mark Beyer, speaking at Gartner’s Master Data Management Summit. “It’s coming to your shores, and it’s challenging everything you do.”
Because the information that is considered big data is usually created outside of your organization, it’s highly unlikely you’ll be able to control its form, Beyer said. New sources of information are created constantly and at varying speeds.
The struggle is to figure out when data assets created by machine sensors, social networks, email or any other sources start to identify a relevant trend.
“In a big data environment, anybody can create a new value at any time, and it grows in persistence and in popularity, and pretty soon it starts to look like master data,” Beyer said.
Value, in this case, means a unique piece of data. But these trends have to be considered only a suggestion for a new value, one that needs to be vetted against current master data. The master data management system needs analysis to understand if the new value is a synonym or homonym, or jargon.
This is where unstructured data analytics methods can actually help; mining text to discover meaning of new words or values, Beyer said. A set of master data can be used like a glossary against a set of unstructured data to link values together, and throw out terms that are misleading.
“I can use master data management to build a structure for managing my big data,” Beyer said. “I’m using what I already know to dive into the next data set.”
The Question of Assigning Values to Data Assets
Big datasets are getting popular because they potentially contain value, if used correctly. This has led to data-heavy companies and industry analysts to consider what actual monetary value data has.
According to national insurance and accounting standards boards the answer is “none.” You can’t file a claim for data loss, and you can’t claim data on your balance sheet.
Gartner analyst Doug Laney thinks that should change; he’s conducting research into a field he calls infonomics to consider how best to value data as an asset.
Hand in hand with that study, Laney said, is the consideration of how master data management would change if data had real monetary value, because “we’re moving beyond the notion that information is just a byproduct to information being a resource,” Laney said.
“Imagine any other asset in your organization where you don’t know it’s cost, and you can’t account for the value it’s generating,” Laney said. “That would be inconceivable to your CFO.”
The result of this thinking will be better management and caretaking of data, to increase its value, and better tracking metrics for how data is used and the value it generates.
“You can use this kind of method to get an idea of the value of all those [analytical] reports,” Laney said.
Home page image of 1939 British government poster via Wikipedia.