Frank Johnson has been taking me to task over my view of MDM in my recent blog.
He thinks that that I'm being rather impracticable as far as smaller companies are concerned "It was enough that these businesses could just get these applications up and running in answer to specific business needs; they had nowhere near the human or financial IT resources... to project a strategic [data management] vision across the entire enterprise".
Well, the MDM tarpit affects lots of companies with all the resources they need to do things properly but Frank may have a point with regard to smaller companies. Nevertheless, I can't help feeling that vision is fairly cheap and an abstracted high-level data model is much smaller and easier to deal with than the physical data structures these smaller companies are presumably already dealing with. The data management issue is real; and I do believe that not managing your data properly and then applying MDM is more expensive (overall) than managing it properly in the first place.
However, there is, I concede, a barrier for small companies hoping to manage their data. The sort of tools which help you do this are often expensive and need highly skilled people to use them. When IT is a part time responsibility for someone in a 3-person company, it's unlikely that there going to be an expert data analyst available.
Perhaps we need intelligent database products which allow their users to work with the data model in business terms and which automate the underlying data analysis and generate the appropriate physical databases? Something similar to what used to be called a 4-GL, perhaps?
To illustrate what I am talking about at the smallest possible scale, my PDA has (as is common) a business-card based contacts application. It suffers from the obvious failings of an un-normalised database—if I delete the business cards of my IBM contacts (say) when they leave IBM, when my last contact leaves IBM, IBM disappears from my contacts database, even though IBM is still there. So, I resort to writing a dummy-contact card to hold IBM-specific data—and now when I add up my list of contacts the number is inaccurate.
No-way should I have to know about normalisation as the user of a PDA contacts application; but all contacts applications are very similar. So how hard would it be for the application to normalise itself in the background, and set up a separate company file for itself, with pointers from a business card image in the presentation layer? It could then ask intelligent questions when adding a new card causes a conflict: "this is a new phone number for IBM, should it replace the number I have or do you want to label it as an additional IBM phone number?". Isn't this a small part of what IBM, for example, probably calls "smart technology", now possible when even a PDA contains a powerful general-purpose computer?
Well, that's perhaps one way for smaller companies to get access to what I'd call good governance of data management, a.k.a. "going it right". But a SaaS company called Intralinks has just reminded me of another way for smaller companies to tap into "best practice". Why not just outsource data management? An external service provider can afford to maintain a data analysis team and could normalise your data, identify duplicate entities etc. and could supply you with a business-level model which helps you visualise, re-factor and reuse your data resources—all as part of the service.
This isn't exactly what Intralinks does but it is a feasible service offering I think. But perhaps you are worried about losing control of your data? This is where Intralinks comes in, as it demonstrates that this need not be a problem. Intralinks offers a SaaS (Software as a Service) solution to managing the distribution and delivery of mission critical and often highly confidential documents involved in mergers and acquisitions, clinical trials, construction projects and so on, often for the largest and most important companies. Part of the service delivered, with no upfront capital cost, is best-practice security, compliance and audit-ability and use of its expertise in managing secure access to information on a "need-to-know" basis from both within and outside your organisation.
Intralinks has been doing this successfully since 1997. If smaller companies are indeed resorting to MDM solutions to address failings in their basic data management because proper data management from the get-go is too hard, is outsourcing your entire data management to an expert SaaS data management provider from the start going to become "best practice" for SME data management in the future?