One of the central tenets of EDM is that it should promote trust and confidence in a firm’s data assets by applying a consistent approach to assuring data quality. However, in order to achieve that goal (particularly with regards to consistency), it is important that EDM teams can service all of their stakeholders’ data requirements—from the front- to the back-office.

Traditionally, however, most EDM teams have tended to focus predominantly on back-office requirements. In part, that may be because back office data requirements are more technically straightforward. They involve datasets that update less frequently (end-of-day prices instead of streaming intra-day) or are less challenging to compute (bond payment schedules instead of yield curve analytics).

Meanwhile, business users in the front office have largely been seduced by the allure of high-end data terminals (with accompanying APIs and Excel add-ins for accessing data outside of those terminals). As such, they have tended to view EDM teams as less responsive to their needs. And they have not shied away from acquiring their own specialist content sets, developing their own analytics and generally operating outside of centralised EDM quality assurance processes—all in the name of being more agile from a business perspective.

This organisational friction between centralised EDM teams and the business units they serve tends to be most evident when it comes to specialised and complex datasets. If there are requirements that EDM teams struggle to support, then business units can justify adopting a more maverick approach.

The solution is simple. EDM teams need to be more in tune with business requirements. They need to operate EDM platforms that are agile, business-ready and collaborative. They need to ensure their business colleagues have no excuse to work around them.

In our opinion, the four most important characteristics that an enterprise data management platform should offer are:

• i) Complete instrument coverage: Offering complete instrument coverage is easier said than done. For a data management platform to be truly business-ready, it will need to capture the nuances of each asset class (eg. business logic and data dependencies) within its data model. Moreover, EDM systems need a data model that is sufficiently flexible and extensible to accommodate ever-evolving financial innovation (new products and structures).

• ii) Support for all derived data and analytics. Proprietary analytics are crucial to any investment firm. Having your own view on pricing, risk, fundamental or technical analysis can give traders and investors an edge over their rivals. However, from an EDM point of view, proprietary analytics can be a real challenge, which doesn’t just involve feeding analytical models with clean underlying data (including historical time-series). It also means integrating the output of those models (even complex data objects like curves, surfaces and cubes) back into the EDM process to be cleansed, validated and made available for distribution.

• iii) A collaborative approach
  Data management ultimately needs to be driven by business requirements. That means EDM teams should adopt a collaborative approach with all of their business stakeholders. All of those stakeholders should have access to the data they need via their choice of interface, whether that be an end-of-day file, snapshot intra-day feed, pub-sub mechanism or direct API integration with their chosen business application. Equally, EDM teams should engage business users in supporting data quality, deploying self-service capabilities that allow them to fix incorrect data points directly (after-all, who better to identify issues with data quality than the business).

• iv) Audit Trails (with regards to data provenance and process workflows)
  When it comes to EDM, the topic of data provenance is very much in vogue. Although one would like to think that the entire enterprise will innately trust a data item that has been promoted to gold copy status, invariably some users will want to know the original source, which validation checks were carried out and who ultimately promoted it to a golden copy. Equally, maintaining process audit trails can be crucial for management information systems – enabling teams to set and enforce a full range of performance targets – from the average time taken to on-board new content, through to resolving and fixing exceptions.

In any large, complex organisation, there will always be friction between centralised IT and data groups, and the business units that they serve. By making sure centralised EDM teams are supported by technology that is agile, business-ready and collaborative – you stand a much better chance of servicing requirements across the whole enterprise, and achieving a consistent approach to data quality.

This blog was first published on Bobsguide.