Do Master data management one piece at a time

Deciding whether to enforce Master data for disparate entities across the enterprise is the $1mill question. Due to immatureness of the organization or shallow business drivers a fractmented approach might be preferable rather than the full deal. You might recall the article on my sight telling about what is the necessary cornerstone’s in a successful MDM initiative, these steps is still necessary, yet they can be addressed in various degrees. Taking basics in a BI project often seems as good grounds to initiate a MDM initiative, while the obvious disadvantages of not having a MDM driven approach to your data is obvious. Rather than repairing with ETL, you might consider a advocating for a MDM initiative. I am sure you will hear a lot of objections, however I been right in my claim “it doesn’t take longer to-do it right the first time” quite a few times.

The +EDM guys sees MDM implemented in such small steps, they tend not to believe in BIG BANG implementations. Moving in.. From a small department with a small given task and then expand from there. Guess its like going up a mountain, there is a few approaches.. There is the steep fast way up, with an increased chance of failure, or the path twisting up.

I guess I fully agree with this, however, chances are your will not be able to find your sponsors high enough in the organization. And the obvious disadvantage of not having your foundation at a level with the “swings and roundabouts” overview, is lack of investment will, both economically politically and resource wise.

AS IS

Importing data in MDM is done through an interface table. It is possible to call a stored procedure that kick off the import of data. This crude mechanism doesn’t support any features in a SOA architecture. This calls for a SSIS integration rather than the desired real-time integration one would prefer with a BizTalk Enterprise hub. In Redmond Microsoft didn’t reveal any plans of changing this for the first release.

 The Secret to Driving ROI in MDM

"Get the Business Involved"

Certain cornerstones are essential when driving a MDM project to success. Mostly the Sequence of these steps is important, but it’s hard to give a general starting point, while it often depends on the tasks at hand. One can compare it with going to the doctor, his approach isn’t quite the same when going for a full medical checkup or when you arrive to the emergency room with an broken leg. So where to start and how much weight you put into each of the topics listed depends on the projects needs, however a full circle is needed before a success can be in sight, and the MDM patient cured :).

Ownership Defined

It is important that named persons are held accountable for the structure of Master Data and quality in data. Ensure that the sponsors has identified the subject matter experts or data stewards across the enterprise who can best articulate what quality means to your business, identify process gaps where improvements are necessary and define the quality metrics that should be monitored ongoing.

having Common Definitions

One of the more hard tasks is getting fractions of the enterprise to agree on consistent definitions on what e.g. a vendor.

What is a product? Is that Product stones, and on what granularity are they?

Defined SOR/SOE

Being able to have an answer to the question: From where do I get Master Data I can trust?  (e.g. a product) across processes and IT systems. There is no single truth which constellation of SOR/SOE there should be chosen, only one rule stands firm, it’s essential that it’s uniquely defined.

I like to emphasize that because MDM is the consolidated truth of how the entity is accepted corporately does this not mean it’s the single point of the facts. It is rather a single version of today's corporately accepted view. Remember: what's true today was wasn’t necessarily yesterday truth. Master Data is auditable as a system of record.

governance defined

Defining governance roles, maintenance processes, data principles, data quality monitoring, procedures for continuous improvement

Lets take the situation where a sales person gets in contact with a new customer. Ungoverned he might be very interested how much money he can squeeze out this client, but less focused on the rest of the procedure.

So he gets the deal down, Company Name, time of delivery, 5 times the red car and 2 of the blue. Price, zero in discount and he is happy.

Then what happens when the order hits production, they got like 6 different red and also the model with motor. When is the earliest and latest date of delivery ?... Hmm they call the customer. Well the contact person is missing so they trouble a number of people in the customers organization before finally getting the info needed, luckily they spoke to him because he orders yet 2 more cars, cool.

 

Then it moves to shipping, where should the cars be delivered, who will pay for insurance etc. again information is lacking, calling the sales person does little good, so calling the customer seems like a sound option. Again after troubling a number of people in the organization they get the information.

 

Now its time to get the big bucks, 7 cars nod a bad thing, hmm weird the production order doesn't match the purchase order. Better call ….

Does this leave an impression of professionalism at the client. Does it affect business?. Do it cost more in man power to be with out governance.

Solving the right things for the business

Define and validate the business need

Identify the specific business processes that need to be improved or enabled with master data and map those processes

both to the systems - upstream and downstream - and the enterprise data that is critical to supporting each process. Focus on the people in the organization who will be benefiting and how, not only on the capabilities delivered.

Sponsorship.

Your MDM initiative must be supported at the highest levels possible, and your executive sponsors should recognize it as a priority investment that will require ongoing oversight and management from both a business and an IT perspective - it’s not a onetime project.

I use to blame Michael Vølund for this method. GET THE PAINPOINTS, once you know where it hurts then you can ask the business to make these pains operational by putting value to them in terms of money and extra effort. But this it quickly turns into business drivers. Then seek sponsor ship as high in the organization as possible. Preferably so high they can see its all “swings and roundabouts”