Been away too long – sorry folks. Very busy here – MDM Hype Cycle work and MDM Magic Quadrant season too. Oh for a simple life. When too busy, blogging falls by the waywide. Or, as in this case, until an emotional string is pulled.
Who would have thought that an innocent suggestion would become a veritable nightmare! Some time ago I was talking with some business intelligence (BI) analysts at Gartner about how BI implementations manage data, and specifically master data. It became clear that there were some technological similarities to “implementing BI” and “implementing MDM”. One difference though is that BI is “downstream of the business transaction/operations applications” and MDM is “upstream, in the business applications”. To highlight the potential leverage point of the common technology elements, I coined the term, “analytical MDM”. Five weeks later users were calling in asking for analytical MDM tools and vendors!
Given that we can all agree (really?) that there is no such thing, literally, as “analytical MDM”, it is interesting to think again about the topic. What might the trends be, for example, on spending on technologies related to analytical MDM?
- Analytical MDM is a “use case” in that it describes where in the business the focus for “MDM” is (downstream, toward BI)
- Analytical MDM represents the use of certain technology for mastering (master) data within a BI/DW environment
- Entities included in scope of analytical MDM are customer, product etc, but also more of a focus on hierarchy and dimensional data
- Governance in this MDM use case is mostly passive (rules and/or transformations) and not active (business users, policy evaluation/revision)
There are two dynamics that challenge this view:
- Few users actually want to “do” analytical MDM in isolation; it is often part of a BI roll out
- More sophisticated users do actually create new data (hierarchies) for business user “what if” views (e.g. new sales territories) and as such, these “BI solutions” start to look less like classic BI applications but more like regular business applications. As such, active governance should be established.
It is this last dynamic that leads to the conclusion: Analytical MDM will:
- Exist as part of BI when BI operates independent of BA, and
- Will become obsolete as (operational) MDM establishes active governance upstream in BA, thus simplifying what BI needs to do with respect to managing master data
I had a real world experience two weeks ago that reinforced this. I was sitting down with a global media/marketing organization. This business was heterogeneous (from an IT perspective) and autonomous (from a business managed perspective). The organization has just spend millions (!) on a global BI roll out. These guys were in year 3 of a 4 year roll out, and they were pulling their hair out. The deal that was signed included a passive (they did not use that word) data governance effort as part of the centralized BI data warehouse where IT would act as steward in order to get to some semblance of “single view” without all the cost and overheard of an operational MDM program in the business. It all looked good; it looked cheaper than the alternatives. It looked like a sensible plan.
I was called in because they now have a problem. The business started to get interested in the master data and MDM program. The business started to spot how global and local data was related, and how the responsiveness of one business assisted or impeded the success of another; that customers were connected and so their businesses needed to work more closely together. The business came to conclude (they did not use these exact words) that they needed to be active participants in “governance” as part of their daily life! The organization was pulling their hair out since they could not figure out how to support an operational MDM program with the tools they were implementing. Bottom line- they can’t. They selected tools and technology for use downstream in BI (they fell for the vendor line) and their “analytical MDM” vendor/technology of choice is ill-suited to support a fully fledged operational MDM program. Ooops.
End of the story: This organization has a lot of re-work ahead of them; IT has lost face; and worse, the cost of the whole thing will now be higher than if they had thought this out properly in the first place.