MDM via Technology: An Incomplete Recipe

Want a recipe for integration failure? I have one for you:

• Embark on a complex data integration project. For the sake of today's meal let's be bold and make it a Master Data Management (MDM) feast.
• Have the IT group research and buy some MDM software. Hire a technology consultant to teach them the intricacies of the software.
• Identify a couple target systems to tie to the MDM app. Give IT a big spoon (some cool ETL software) to stir with. Cover the pot and let it simmer.
• Separate the flavors by keeping business and IT from interacting.
• After a couple months, lift the lid and add a dash of "when will you be done"?
• In a couple more months, check the pot again -- watch the steam patterns as IT grumbles about the data not matching up.
• Add a dash of "be sure not to break any of our processes" and mention we don't have time to consider business changes.
• Watch the steam during unit test.
• Savor the pungent flavor of an unhappy user acceptance test.

Where does the recipe go wrong? When it treats Master Data Management -- or any complex integration iniative -- like a technology project.

Loraine Lawson rightly points out today that today's complex data integration projects are business initiatives first and foremost.

Data Integration is no longer a matter of techies pushing data from point A to point B.

The Ideal Meal

In Master Data Management we, the business as a whole, are charged with pulling together cleansed, high-quality data from points A and B and exposing the result to C -- in a way that makes sense to C's core business logic.

To do this, we have to

• match up the business rules, codes, and processes of A, B, and C in a sane way,
• know who to turn to for rulings when data sanity breaks,
• and be ready to integrate process "D" into the whole shebang at a moment's notice.

Necessary Ingredients

Data Integration and MDM are BUSINESS ISSUES. Technology is just the enabling glue.

Marcia Kaufman rightfully points out that companies historically have managed information in a siloed way.

This siloed approach (and the IT-centric implementation methodology that enabled it) are not valid in the world of SOA, MDM, and multi-system integration. Rick Sherman at Informatica echoes this in his post title "MDM is Not a Product" -- discussing a Forrester Research report on MDM. (As employee of a company that sells MDM software, I would temper Rick's statement by saying "MDM Solutions Require More than a Product.")

What is the Recipe?

The short answer is to incorporate business process and data governance into the solution. I've helped build out my Practice's methodology for MDM and Data Integration, and we incorporate "people and process" components at all points.

It's the responsible thing to do, and the family feast tastes a whole lot better.