The master data management (MDM) industry is, depending on your view, at least 10 years old now. As with any young child, it stumbled about in its early years, uncertain of its direction, but has now started to mature a little. Initially we had products aimed exclusively at either customer (CDI) or product (PIM) data. Then around 2003 the first multi-domain MDM products appeared, and for a time there was some jostling for position between the domain-specific and multi-domain offerings. A flurry of acquisitions saw many of the pioneering companies swallowed up by larger vendors as the market became fashionable. Within a few years IBM, Oracle, SAP and Microsoft all had MDM offerings, as did integration vendors such as Informatica, TIBCO , Software AG and Talend, as do SAS and others.

Looking at the landscape today, the dust has started to settle. The large vendors have their offerings, and are seeking to position these in the context of broader data management "platforms". Along with an MDM hub they generally bundle some data quality component (sometimes itself licensed from another vendor) and the promise at least of integration with this and their data integration offering e.g. an ETL technology. Some support for data governance administration is usually present to one degree or another. With this message the larger vendors seek to shut out the remaining pure play specialist MDM vendors, who mostly focus on product data. Why go to the trouble of evaluating and dealing with a new vendor when you can just plug in another software offering from a familiar incumbent vendor?

This can be a compelling argument to end user companies wanting to simplify their spaghetti-like software landscape, but purchasers should be careful before buying into such a comforting vision. For a start, the degree of integration between the MDM, data quality and data integration offerings of the mega vendors varies considerably. Some have done a good job of linking their products together; SAS for example, has mostly coded from the ground up, so has less integration work to do. Others have gobbled up multiple MDM offerings, and are still trying to digest these and get them to talk to one another in a meaningful way, never mind seamlessly integrating data quality and data integration into the mix.

From a buyer's perspective there is always a trade-off between a broad platform and best of breed. If you buy the best possible MDM hub, data quality tool and integration product for your needs, then does the functionality advantage of this balance the additional integration work needed? Would it be better to buy a less-than-optimal combined suite, which at least in principle should work together more easily? There are no simple answers here, as it depends on the individual use case: if the key business requirement for you is (say) highly accurate, fast and flexible record matching, then the advantages of picking the absolutely best specialist product may well dwarf the additional integration effort required to get this talking to your current ETL tool. As noted, the degree of integration of the components from the larger vendors varies dramatically, with some of the integration more apparent on PowerPoint diagrams and vision statements than in actual software.

Moreover, buying from industry giants is rarely as "safe" as it appears: they may be less committed to a particular product line than a specialist vendor, for all their financial resources. For example SAP started with their own MDM development, dropped this in favour of an acquisition (of A2i) that became SAP MDM, and have recently switched horses once more to another in-house development route with MDG, and have recently purchased hybris too. Oracle has multiple MDM offerings, and IBM has still to fully integrate the three separate MDM products that it has acquired. By comparison, some smaller specialist vendors have been crystal clear in their direction, even if they do not have the bank balances and global reach of the industry giants.

The MDM industry is maturing, but it is still a long way from the stage where MDM products are commoditised. There are significant functional, scalability and performance differences between the offerings, and purchasers need to carefully articulate their precise requirements before being in a position to make an optimal decision as to whether to go down the platform of best of breed path. They need to test out the software choices thoroughly on their own data and ensure that their specific requirements are met, and not assume that vendors' claims are true. This is even more the issue when looking at roadmap plans from vendors, which may take much longer to materialise than the vendors claim. The MDM field has fewer independent players in it now than a few years ago, but buyers still face a complex choice of both strategic direction and satisfying their immediate needs.