2004 will be FAME's 21st year in the data management business, and our first as an operating unit of SunGard (NYSE: SDS). The intervening period has witnessed huge changes in the way financial institutions collect, manage and distribute their data, with recent years recording some of the most dramatic and interesting changes.
Daily discussion with hundreds of existing and potential clients has provided us with valuable insight into the data management concerns and issues facing today's financial institutions. Many have argued that the current mantra of the "Golden Copy" is far from being the solution to many of their data management problems. If not properly understood or implemented, a "Golden Copy" model is likely to become an integral part of the problem itself, and has the potential to be a major factor in project failure.
We believe that the "Golden Copy" is still the answer, but not in the manner that most have come to define it. To understand our approach, one must first to examine the nature of the industry.
For a variety of historic reasons, the management of market and reference data evolved into a very decentralized model. Most financial institutions today operate a myriad of disparate systems serving a variety of separate business silos, divided along product, geographic and/or business process lines. Typically, each silo is concerned only with data of direct relevance to its own operations. As a result, much of the data, and the entire effort of data management, remains duplicated across the organization as a whole.
This decentralized model naturally leads to cases of inconsistent and incomplete data, both across an organization's internal systems as well as within the wider investment community. It also raises efficiency and cost concerns, as the duplication of maintenance, hardware, and data subscription fees can drastically affect the bottom line. The fact remains, however, that data is a key component of many operations. As such, each business silo tends to jealously guard its own data management operations, and is often unwilling to cede control of something, even to another unit of their own organization, that is so critical to the success of their business.
On the face of it, the solution is simple: replace the multitude of decentralized data management solutions with a single, centralized "Golden Copy" database... one that will, in turn, feed all dependent systems with the quality-assured, centrally-approved and expertly-managed data set. With enough buy-in from senior management, localized resistance can be overcome, and we can all drink from the holy grail of clean, consistent market and reference data. This, in a nutshell, is the current "Golden Copy" mantra.
Not Another Database!
Most financial institutions already have at least one security master file, and maintain a host of databases and a multitude of complex data models. Adding another large database, with yet another complex data model mapped to all the others, will only introduce more complexity and open up more potential points of failure. Clearly, the problem is not the lack of a database, but rather the data itself held within these databases. The solution is a process, not a database.
SunGard's referencePoint™ solution addresses this issue in a bold and innovative way; it does not impose a complex data model. referencePoint represents all data in a flat structure, allowing it to be easily mapped and exported to the existing data management systems of an organization. In other words, referencePoint works "with the grain" of your current environment, rather than ripping it out to replace it with a more complex alternative.
Using referencePoint, an institution can put in place a centralized control process to deliver cleansed and consistent data throughout the enterprise, without having to radically change the data management landscape of the organization. This provides huge savings both in project implementation time and cost, while simultaneously removing one of the largest and most common reasons for project failure.
So whose "Golden Copy" are we going to use?
Many centralized market and reference data management projects fail to meet expectations due to politics. Getting all of a large financial institution's many data consumers, with their widely disparate requirements, to agree on what should be included in the "Golden Copy" database, how it should be sourced, cleansed, and corrected, and when it should be delivered, is such a Herculean task that many fear it is a practical impossibility. Detractors will argue that compromises will have to be made, or someone's requirements will have to be left out. Either way, they argue, some end users will not be satisfied, and as a result, will rely on their own initiatives to provide a solution, bringing us back to square one -- decentralized data management.
The problem lies with the basic assumption that there is such a thing as a singular "Golden Copy." Many will argue that each internal consumer of data has their own definition. For the most part, requirements will overlap (since after all, a SEDOL is a SEDOL and an ISIN is an ISIN). However, items such as the price, from whom it is supplied, how it is calculated, when it is taken, etc. will vary greatly across an organization's many departments. Clearly, a one-size-fits-all definition of "Golden Copy" is not realistic.
The reality is that there will be a "Core Golden Copy," common to all consumers of data, such as instrument codes and most static/reference data. This could be as much as 60 to 80 percent of the overall data required. The devil in the detail, however, will be the remaining 20 to 40 percent that varies from consumer to consumer. This "Consumer-specific Golden Copy" is where many projects start to get into trouble, and without an adequate answer, most projects will fail. Of course, any viable solution must capitalize on efficiencies of scale. The "Core Golden Copy" must be sourced and managed only once, to avoid redundant costs and the duplicated expenditure of resources. In other words, consumers cannot have their own, individually distinct databases.
This is a concept that we at SunGard call the "Multiple Golden Copy." As we mentioned earlier, the lack of a rigidly enforced data model allows the flexibility to provide the ideal solution. Each consumer of data can have their own unique and distinct definition for how data should be sourced, cleansed, derived and delivered. This can be easily accommodated into a framework where data that is common across consumers is only managed once.
This point is critical for large financial institutions serving many internal consumers. However, it becomes absolutely paramount for service providers such as custodians, stock exchanges, and other data providers, since a managed data offering is only financially viable when efficiencies of scale can be achieved through the "Multiple Golden Copy" concept described above. Indeed, "Multiple Golden Copy" is at the heart of SunGard's own outsourced data management offering, referencePoint Plus.
The result of a successful solution, for both end-users and senior management, is clean, consistent data across the enterprise, sourced once (to keep vendor costs to a minimum) and cleansed centrally (to avoid the duplication of effort and reduce maintenance costs). This will only be achieved with a solution that places the emphasis on a flexible data management process, rather than monolithic, rigid software systems. A solution that cannot effortlessly map onto an organization's existing landscape, while simultaneously servicing the varied requirements of internal consumers, is doomed to failure.
Built on FAME's 21 years of experience with the world's most demanding financial institutions, referencePoint Plus is uniquely positioned to answer these challenges. And now as an integral part of SunGard, one of the leading providers of integrated solutions for financial services, our customers have even greater access to a wide array of complimentary, best-of-breed products from a solid technology partner.
888 7th Avenue, 12th Floor
New York, NY
Contact: Neil Edelstein