saint valentine

MDM without data governance is ‘data integration’, not MDM!

by Andrew White | March 21, 2012 | Submit a Comment

Blogs would be boring if they didn’t incite a little riot now and then. So here’s my effort.

I was reading a recent article on Information-Management, titled “Data Governance and MDM more Successful Together” that reported findings from a survey executed by The Information Difference. I have a number of issues and thoughts regarding the article, and therefore the survey. Bearing in mind that survey’s are dodgy at best; there is an art and a science to getting real data other than interpreted data, and even we struggle to get something from survey’s that is truly insightful.

First up the title of the article. MDM contains a major element (dare I say, building block?) that is about governance. MDM includes governance – it is about governing (for better business value) master data. So you cannot actually “do” MDM without a governance component. On the other hand, “data governance” does not actually tell you much at all other than the fact that you plan to govern some data. What data, you may ask? So arguably “data governance” is very broad and not very specific, and MDM is “data governance applied to master data”.

Second, the first sentence of the article: “Master data management has moved from being the new kid on the block to now be recognized as a mature and accepted technology widely regarded as essential for delivering reliable business data.” MDM is not a technology! It is a discipline, which is operationalized by some technology(s). I would accept that various technologies that used to support an MDM program are pretty mature (think data quality, data integration) but many aspects of an MDM program remain elusive, complex, and a right pain (such as establishing stewardship roles in the business and setting up the operational tools to support the “day in the life of the business data stewards”.)

The third sentence is key – because it exposes the “flaw” in how users responded to the survey, or how the survey seeker ended up drawing conclusions based in a misunderstanding on what people say they do, and what they do. It says, “There is scant information available on the approach being adopted by organizations that have implemented both data governance and MDM.” This is a great point; it means that those users that responded to the survey were not implementing MDM at all but most likely where implementing ‘yet another data integration effort’ to tackle the same problem! With the lack of governance and day to day business oriented data stewardship, MDM does not take place and IT is left doing all the usual tricks with technology. Thus the root problem, weakly governed business processes, remains in place.

The next paragraph also implies a difference in how we look at MDM and how we talk with our customers, and how the survey did. There is reference to, “In particular we wanted to get feedback from respondents who have both data governance activities and MDM projects, to see to what extent the two really are linked.” Well, MDM is not a project! MDM is a program, that will spawn off and launch all manner of projects, that span and include DQ work, DI work, deployment of a hub, data modeling, and the list goes on. None of these individually “are” MDM. But when organized in a specific way, oriented around a discipline to change the way in which the business users crate, use and abuse their own data, and then MDM persists.

Some of the findings, reported in the article, were quite interesting. Apparently data governance programs and MDM programs (better term used later in the article) started around the same time. I don’t quite get that. Would these DG programs be oriented around non master data? Or part of MDM?

The most exciting finding I spotted was this: “Encouragingly, for the majority (63 percent) of data governance programs, the scope and context includes MDM, but is in many cases much broader. So, in most cases data governance and MDM are closely intertwined.” This is exciting because it explores the reality of MDM. Though we talk of it as a discipline, established as a program, spawning projects, the reality is that MDM never stops! No user we have seen that is successful with MDM (that is, they are accounting for improved business outcomes) stops with MDM. The scope of information that is governed for business value increases over time. As such MDM is poorly described or perceived as a discrete amount of work. It is actually a commitment to change how information is used, starting with master data. MDM is just the first of many steps. If you stop at MDM, you will start to go “backwards” in time and most likely data consistency will fall and business outcomes or integrity will be no less assured. But if you are smart, you will build ever increasing “building blocks”, one atop the other, as you evolve from MDM towards an enterprise wide Enterprise Information Management strategy. And that will take years.

The point of the survey is worthy, and the analysis and reporting of that findings well worth it. But it is so difficult to get information from such things that help with the next decision. I would much prefer to highlight how those programs, targeted at MDM (sustaining single version of the truth in order to maximize business integrity) do better when their have adequate governance support/focus. With too little, or non, the program ends up not being MDM at all.

DESARROLLO

1. Reading = Lectura

2. Recent = Reciente

3. Information = Informacion

4. Reported = Informar

5. executed = Ejecutar

6. Difference = Diferencia

7. Interpreted = Interpretacion

8. Article = Articulo

9. Element = Elemento

10. Component = Componente

11. Actually = Realidad