Good Things Are Usually Expensive: MDM Case In Point

This blog entry originally appeared on the ebizQ blog, "Data at Your Service" Feb. 10, 2010.

In the past few weeks, the two largest independent Master Data Management (MDM) vendors, Initiate and Siperian, were acquired by IBM and Informatica. Inside any company that is big enough to run their business on application software (packaged or internally built) and not on Microsoft Office tools, there is typically a problem of how to get a "single view" of the customer, the supplier, the products, or some other business pertinent entity. For example, fragments of customer data might be spread over the billing system ("the" accurate address), the CRM system ("the" accurate contact number), the marketing automation system ("the" accurate email address), the customer support system, and other more obscure places that the company execs would cringe at if they knew.

MDM promises to tackle the unchartered mess of your data center and provide an easy to use GPS to your most important assets. Once the MDM is deployed, any application can quickly and easily get to the key information necessary to run cross-sell and up-sell campaigns, to fix the very annoying problem of customer support centers asking for your information time and time again while getting transferred throughout their call center, to provide executive management a real time dashboard of the business, etc. As I stated in my last blog entry, data services is a collection of data integration software that increases the speed and agility through which we can get to and process our most meaningful information. MDM is certainly an important type of data services software ... to some companies.

If MDM delivers on the promise above, why doesn't every company have one of these? Let's not stop at one, give me four or five. MDMs for everyone. Yes, the benefits are great, but what's the downside? Any large business CIO knows that solving the data-integration mess inside their company is going to be expensive - really really expensive. Not necessarily because the costs of the MDM software is overly expensive, but because using an MDM requires the company to know a lot about their data, i.e. where all the fragments of data are located (how many acquisitions and retired systems do you have?); how to get to the information ("what's an OCCURS DEPENDING ON again?"); how to figure out which one of the twelve customer names you have is actually the correct one, etc. MDM software does not typically come bundled with data integration software - that exercise is left to the reader. That is, successful use of an MDM system assumes that you can or have wired together the important fragments of information so that the MDM software can use it. In reality, what that means is that you're going to spend millions of dollars doing data integration in conjunction with using the MDM system. Once the data integration work is done, MDM provides robust tools and processes for data matching, data quality, data cleansing, and data delivery that will ultimately provide the "golden" data to the businesses.

What's a realistic expectation for how much this would cost? To solve a single view problem for a very specific piece of your business could cost between $500k and $2m in MDM software fees plus 4-5 times that much in consulting to get the system going and this holds if, and only if, you already have a good handle on your data. It's not atypical for large companies to use MDM at a departmental level because solving a pain like cross-sell/up-sell might be well worth the $5m it costs to make it happen for that very focused part of the business. So, yes, I could end up with multiple MDMs throughout my company. And, yes, if my business is decades old with all the problems I mentioned above times 10 then I could easily spend $50m to solve just the single view of the customer pain - this is not an exaggeration but might well be worth the $50m if it means I can see another $1b in revenue that the investment will drive.

MDM can provide real-time access to a single view of information inside your business and is a key option to evaluate when looking at data services to provide competitive business value; however, set expectations appropriately on the economics. Next week we'll look at alternative data services technologies for solving the single view problem.