Key Components of a Metadata Program

Ready to Develop a Metadata Program?

In Previous Articles in this Series we have Covered:

• Why are we not leveraging our metadata today?
• Why projects fail.
• Common perceptions that prevent us from utilizing our metadata.
• How and why metadata management is important.
• That business metadata holds the key (focus should be here).
• That technical metadata is best leveraged for searching and impact analysis.
• Business terms, business rules, and taxonomies are building blocks.
• Metadata entry and stewardship is part of the required tasks.

First lets have a look at some basic components of a typical metadata management program:

• Defining the problem to be solved.
• Determining initial scope and requirements.
• Getting a budget and executive sponsorship.
• Knowing where the business metadata and technical metadata fit in.
• Knowing your options buy vs build or hybrid.
• Project execution.

Common Problems Solved with Effective Metadata Management:

• Can’t reliably search information from other programs.
• Terminology often means different things in different contexts.
• Information from other programs may be contradictory,  e.g. research information may conflict with regulations.
• No glossary business terms.
• No data dictionaries exist.
• No way of knowing how things inter-relate.
• How is a particular item calculated?
• Redundancy of effort.
• Duplicate data, or poor data quality.
• Lost knowledge due to employee turnover.
• Cannot identify information for compliance, audits, and governance.
• What is the impact of changes to a data point? Downstream or upstream.
• Is there a single version of the truth? Can I trust this data?
• Employees spend too much time searching for information to do their jobs.
• Metadata management is required for master data, data governance, data quality, and big data efforts.

Common Mistakes

• Often we want a magic bullet to reverse engineer sophisticated products that their own vendors cannot provide useful metadata for.
• Metadata ETL hell. (Gory details – leave that to ETL Tools).
• Tendency to get bogged down in details, theories, and architectures  (leave theories to the academics).
• Analysis paralysis.
• Scope too huge. Try to sell metadata mega project ( RFP a good example).
• Expectation or belief that standards actual matter.
• Opportunistic vendors. I.E:  BI tool vendor claim they are metadata management tool.
• ETL is part of the problem , not the solution.
• No budget or buy-in.

Determining Initial Scope and Requirements

In order for any project to be successful, the better job we do up front of defining requirements and scope; the more likely we’ll be successful. I cannot emphasize this enough.

Getting Executive Sponsorship and Budget

This is key, for obvious reasons; without budget and buy in there will be no project. In order to provide a good metadata return on investment to the decision makers, you will need to leverage the requirements gathered to identify the key pain points and drive the message home as it pertains to the problem you will be solving.

This is the most difficult part of your effort, as it will involve getting some hard numbers and building a clear presentation to management in language they understand. Typically, technical folks have difficulty with this and usually need to ask for help in this area.

Knowing Where the Business Metadata and Technical Metadata Fit.

I have included this, because I’ve seen way too many efforts get unwieldy because there was simply too much focus put on the technical and the reverse engineering activities. We need to know where the real benefit is. Propose a phased solution that targets the business need, which means we’ll be focusing on business metadata. You’ll get a win.

Buy vs Build

Here, I have to say that if you expect to build a metadata solution from scratch, you are in for a lot of pain. Take it from me, unless you have a lot of developers and resources to spend the time it will take to go through full blown development projects, bugs, and iterations; you are doomed to fail.

Unbiased as I am not. I recommend that you are much better off finding a vendor to work with as a consulting partner, that can help you with requirements, developing a road map, and ROI for your program. Take advantage of commercially available tools that are a fit to your companies needs. It’s not only about tools or technologies, standards or hype, it’s about solving business problems.

Ultimately, the key to your success will hinge on your willingness to get the help you need to develop a program that will solve business problems, and get a win with your metadata management program. The tools will support your effort.

Project Execution

All successful projects have one thing in common, good project management. In order to meet the budget and objectives, the project needs to be expertly managed by seasoned professionals, especially, when dealing with the dynamics related to metadata integration and management programs. However, there are many success stories, case studies and proven ROI samples available to back this up.

Here are Some Lessons Learned

• Must manage expectations and scope. If you try to boil the Atlantic ocean you will get burned. Scope creep is a killer of any IT project.
• Because the term metadata is abstract and not widely understood in a corporate environment, asking someone “How they use metadata” in their job is  a question many people struggle with.
• Trying to define an elaborate set of roles and in doing so introducing a whole new strain of resource requirements and terminologies. Don’t over-complicate.
• Metadata is only useful if it is current and accurate.
• Keep it simple (to start).
• Identify the (a) problem you are trying to solve , then work from there.
• Phased approach.
• The whole idea is to know what you have and where it is!

Summary

Your success is dependent on identifying a business problem, finding a solution to that problem, and selling that solution to management. Followed up by a well defined and managed project leveraging a metadata integration framework.