CHICAGO -- The most successful enterprise information management (EIM) programs are characterized by close adherence to best practices and a refusal to cut corners, according to David Marco, a well-known data management consultant, author and the president of Hinsdale, Ill.-based Enterprise Warehousing Solutions Inc.
EIM programs encompass several data management disciplines, including master data management (MDM), data governance and data quality, and they require close collaboration among business users, the owners of the data, and technologists, the folks who use their skills to make EIM possible. The goal of EIM is to reduce data redundancy and erroneous information and, as a result, improve everything from the customer experience to business intelligence (BI) reporting capabilities.
Marco, the author of well-received books such as Universal Meta Data Models and Building and Managing the Meta Data Repository, has helped both large and small firms set up comprehensive EIM programs -- and he's seen organizations repeat the same EIM mistakes over and over again.
Speaking to a crowded room yesterday at the Enterprise Data World conference here, Marco laid out what he sees as the top 10 most common mistakes to avoid when implementing and maintaining an EIM program. This is what he had to say:
Enterprise information management programs, by definition, are enterprise-level initiatives that should touch every aspect of an organization, according to Marco. EIM programs may comprise several point or department-level projects designed to deliver incremental value over time, but each component of the program must be designed with enterprise-wide goals in mind.
A silo can be defined as something that is disconnected from the rest of the enterprise. And if an organization is building an EIM program around just one application or just one line of business, then chances are they are building a silo.
"One thing that is always a concern is when you see an EIM initiative run as a standalone project," Marco said. "It's not supposed to be something that five people go off and try to hammer out. It's something that should have enterprise exposure at an enterprise level and, unfortunately, this doesn't happen as often as it should."
Failing to balance goals
Organizations taking on an EIM program need to balance long-term strategic goals with short-term tactical goals. This can be difficult because executives like to see results. One way to deliver incremental results, according to Marco, is to design the EIM program with this question in mind: What are some things that we can do in the next year that have tactical value but also move us down the organization's strategic path?
"Identify those optimum projects that tactically benefit the company but that also move the strategic long-term vision forward," Marco said. "And again, please don't make the mistake of building siloed short-term applications."
Boiling the ocean
Remember to take an incremental approach to EIM that delivers value over time. Telling company executives that the EIM program will deliver results in four years simply won't cut it, Marco said. Organizations need to deliver value earlier in the lifecycle.
"I'm just a big believer in the iterative approach," Marco said. "I believe in three- to six-month project lifecycles."
Organizations embarking on EIM for the first time and government agencies may need to extend that initial time-to-value to about six to 12 months. "In the commercial world, you're trying to find one or two people who can say 'yes,'" he said. "In the federal world, there are, like, 1,000 people who can say 'no.'"
Forgetting the 80/20 rule
Organizations taking on an EIM program often do so under pretense that they need to "go after" every scrap of data in the company. But this is a mistake.
According to Marco, about 80% of the value of information comes from about 20% of the data housed within an organization, so companies should identify and focus their EIM efforts on the most important data first.
"Worry about managing the key data in the organization because that is going to give you the majority of the benefit from EIM efforts," he said
Not ensuring ongoing adherence to maintenance
Taken as a whole, EIM should not be looked at as a project with an end in sight. Rather, Marco said, EIM is an ongoing initiative that requires ongoing maintenance -- a point that many organizations seem to forget.
"A lot of times organizations focus so heavily on initially implementing their EIM effort that they forget [they] need to make sure that people adhere to the rules [they] create," he said. "Embed into your project development lifecycle as many of those EIM processes that you can."
Marco added that it's a good idea to let data stewards -- the business users charged with enforcing EIM policies -- know that their bonuses, raises and other incentives will reflect how well they encouraged ongoing maintenance and adherence to EIM rules.
No active involvement by the business side
Business users understand the value of the information they create better than anyone, and therefore they must be closely involved in EIM efforts from the beginning.
"I always know an EIM effort is heading for rocky waters when I [first visit a company and] meet a bunch of technologists," Marco said. "I need some people from the business, especially in [the area of] governance. That should be predominantly [business-driven]."
Neglecting metadata management
Metadata gives context to information. A retail establishment's product code for a can of Coca-Cola represents one piece of data. But information about the store where that can of soda came from and the town where that store resides are examples of context, or metadata. Without properly managed metadata, individual pieces of information can be largely useless.
"Metadata management and governance [efforts] need to begin before you begin your [large-scale] EIM work," Marco said.
Failing to change the status quo
Organizations need to clearly identify the problems that EIM can fix. There's no point in going on with an EIM effort if a company is still going to have a four-fold redundancy in data, Marco said. The projects must be designed to deliver real changes within a set amount of time.
"We're not doing EIM so that 5% of our data is in error," Marco said. "We're doing it to make things better. Our goal needs to be to give us a competitive differentiator from the marketplace."
Thinking that EIM is easy
Conference attendee Lisa Jane Bonamo, a data systems developer for Chicago-based insurance provider Health Care Service Corporation (HCSC), understands that enterprise information management efforts are far from easy.
HCSC worked with Marco and Enterprise Warehousing Solutions to implement an EIM program that, among other things, is helping the organization manage its metadata environment and couple that environment with HCSC's enterprise logical data model and content management-related efforts. In doing so, Bonamo got to experience first-hand just how thorny an EIM project can be.
"It's difficult because there are so many complexities that you have to deal with to get that information into your tool correctly. You’re not just dealing with one application and [you're] talking with many different people," Bonamo said. "You have the IT people who have their own way of doing things from a technical view, and you're dealing with the governance people who are dealing with things from a personal point of view. You have to merge those things together and [it's] like a marriage. You have to make sure that everyone gets a turn at the table and everyone gets an opportunity to voice their opinions."
Running EIM initiatives like a science project
Junior high school science projects can be fun, but they're usually created to demonstrate specific principles instead of delivering new and useful results. Running EIM projects in a similar fashion is a big mistake. Instead, EIM projects should move forward with a specific goal and tangible results in mind.
"I don't want to hear that an EIM organization exists because [the organization] wants to use MDM principles for a holistic enterprise approach," Marco said. "No. It should exist to solve specific business problems. Our data is in 7% error and we want to reduce it to 2%. We have 8,000 processes within our company [and we] want to reduce that to 1,200. We want to be more efficient. We want to reduce our IT budget. We want to improve marketing campaigns."