I gotta admit: I have serious reservations about master data management.

Oh, certainly I see the need for master data management. Of course you need to resolve conflicts and create a single view of your most critical data. It makes sense.

But that's not exactly a new thing. No, my concern comes from the numerous caveats and warnings I see about MDM implementations gone wrong. Comments such as this remark by Mark Smith, Ventana Research CEO & EVP Research:

“I have seen a lot of discussion and shelving of MDM projects due to the clear lack of strategy and planning.”

Of course, he offers advice for how to avoid this fate, but still. Statements like that give you pause.

MDM is complicated and expensive stuff – there's just no question about that. If it weren't complicated, then we'd all have a single view of our customers and products and it'd be great.

But that's not the case. Instead, there are silos of data and confusion – a problem David Linthicum believes will only worsen as more companies turn to cloud computing:

“We got here through years of managing-by-magazine, and dragging in whatever business systems seemed cool at the time without regard for architecture or MDM. Thus, we've ended up with many stovepipe applications that have their own independent processes and the own independent data. I suspect that the movement towards cloud computing will mean that this problem actually gets worse as more systems are outsourced to cloud providers without a data integration and MDM strategy.”

Ideally, as more people explore MDM as a discipline and a technology, we'll see more clarification about what MDM means, and how vendors are implementing it, and develop clear, best practices for successful MDM projects.

But that doesn't always happen. Witness what happened with SOA which, to this very day, generates heated arguments about what is and is not SOA.

I can't help but wonder if MDM is headed down the same road.

Most recently, I noticed Gartner analyst Andrew White had written a post on whether you could do master data management without buying an MDM solution.

The good news: You can. In fact, last year he encountered three end-user organizations that had successful MDM disciplines, without buying “externally sourced MDM technology.” The bad news: You probably won't. He said a mere 5 to 10 percent of those he meets have tried to build their own MDM, and almost all abandoned the effort, reporting that it was too difficult and costly.

Okay, fine. You're probably better off investing in a tool as part of your MDM strategy.

Good luck sorting that out. As I wrote earlier this month, sorting through the various MDM flavors and approaches can be confusing. Several readers added helpful comments to try to shed a little more light on the topic.

But I abandoned all hope when I saw White's post, and learned that a vendor is developing a unique master data management tool that's basically dependent upon an enterprise service bus. Even White found this confounding:

“The vendor – whose name shall remain anonymous, shared their strategy with us, and I have to say I was more confused at the end of the briefing than I was at the start. ... The vendor has decided to build their own MDM capability – but it is not like any other MDM strategy I have seen.”

I'm not sure how much hope there is for the rest of us if an analyst who devotes all his time to figuring out MDM offerings is confused.

Unfortunately, many organizations may have no option but to muddle through.