Modern corporations are becoming more and more dependent on data to operate. In some industries, data pretty much...
is the business, whether it's digital content, such as movies and music; social media interactions; or customer surveys and reviews.
The finance industry is increasingly moving online -- only around 8% of the world's money now exists in the form of cash. Even traditional industries are becoming increasingly digital. Automobiles are now essentially computers with an engine attached, as are airplanes -- a modern airliner generates several terabytes of data per flight. But with so much data out there, who is responsible for managing it? A survey of more than 1,000 businesses published by software and services provider Experian Data Quality earlier this year found that 84% of the companies still regard data management as either solely or primarily the responsibility of the IT department.
In most organizations, devolving the ownership of data to IT without a coordinated data governance program creates issues, as different business units generally own the budgets and, hence, the power. In a global company, there can be tensions over data between headquarters, regional offices and operating companies, and between corporate departments, such as sales and marketing or production and logistics.
Each business unit wants to have control of the data that it finds important and is unconcerned about data that doesn't directly affect it. A finance organization cares a lot about the credit rating of a customer, but the logistics department cares mostly about its delivery address. Marketing is concerned about the characteristics of customers and which segments they fall into, while sales worries about who controls the purchasing budget and can make buying decisions.
How silos affect ownership of data
Absent an effective governance process, this understandable focus on the immediate data needs of day-to-day operations means data quality and completeness get ignored. For example, a salesman may create a new customer record for a prospect without checking to see whether one already exists, resulting in duplicate and potentially inconsistent data entries in different systems. Or he may not bother to collect additional information about a customer that marketing would love to have because it has no direct impact on his sales targets and commissions.
Worryingly, this problem extends outside the business world, with 10% or more of medical records in U.S. hospitals estimated to be duplicated, according to data cited in an article in the September 2018 issue of Journal of AHIMA, a magazine published by the American Health Information Management Association.
The effect of all this is that ungoverned data ends up in independent and often incompatible data silos, jealously guarded by each business unit or department. An IT department that points out that some data held by marketing is partly duplicated over in sales, or different in the two systems, rarely has the power to do anything about it.
Nonetheless, this silo mentality has been the driver of many major IT initiatives over the years. ERP was supposed to sweep aside the data silos, but it only created a large silo of its own in many organizations. Data warehouses were created to alleviate the problem, but they became hard to keep up to date in the face of business changes. And ambitious master data management initiatives floundered on the jagged rocks of internal politics and the reluctance of separate groups to relinquish control of what they consider their data to someone else.
This article is part of
These data silos matter because an organization can struggle to get a picture of its overall operations. Which customers are most -- or least -- profitable? Do we really need all these suppliers? Which partners and channels really add value? To answer these questions, you need consistent, trustworthy data across the enterprise -- or else decisions are reduced to guesswork.
Combining data ownership and governance
Some organizations have tried to address the issue at its core by reassigning ownership of data back to the business units, but through data governance initiatives that involve representatives from each unit. Data governance committees attempt to create universal data definitions, data governance policies and lines of management demarcation around common data domains, such as customers, products, assets and suppliers. They also assign responsibility for data quality and policy compliance to specific data stewards, in many cases within the business units themselves.
The discussions that result can be thorny, as inevitably, the solution involves giving up power and control. I recall facilitating one such session at a global corporation some years ago and hearing 11 different country managers presenting to their colleagues, one after the other, on why the rest of the company should change its ways and adopt their particular data classification system.
Despite the inevitable difficulties, businesses are gradually, one step at a time, reclaiming data ownership from the IT department, which never really wanted it in the first place. Technology can help with that: New software products aimed at business users rather than technologists have recently appeared that enable self-service data stewardship, data preparation, data quality and analytics and also support data transparency.
It helps that the current generation of business managers and workers is much more IT literate than their counterparts not so long ago. People who grew up with the internet, mobile phone apps and video games aren't fazed by a data transformation workflow diagram.
Collaborate for effective data management
There's no magic bullet, however. Designing a high-quality data model is hard, even for data management professionals, and although it's a good thing that business folks are now more involved in managing their data, some assistance and guidance is still required.
Despite the difficulties, getting the business to take ownership of data is a positive step. Indeed, it is the only way to get to the heart of the data management problems that still bedevil almost every large organization decades after the first computer applications were rolled out.
Data management initiatives led by the IT department will inevitably stumble unless it works hand-in-hand with the business users who create and update the data. Without that, IT teams trying to improve data quality and root out data duplication are essentially playing an endless game of whack a mole, fixing one data quality or consistency issue only for another to quickly pop up elsewhere. Only by agreeing to corporate-wide data governance with responsibility by business units will the foundations be laid for successful data management across the enterprise.