Learning from Enterprise Data Management Implementations

Introduction:

We witness a Paradigm shift from traditional Data Management to Data Governance across the Financial Services world. This is driven by many factors that include the regulatory directions through the ever present cost containment efforts to prim the bottom lime.

The advent of EDM (Enterprise Data Management) platforms have been a key catalyst in the equation for the shift to Data Governance structures. However, the road to EDM implementation is not as easy and straight forward as one would like it to be.

Let me share some of my observations and insights on this journey.

Data Usage within the Organization:

Be it a bank or a financial services institution of any scale, the typical drivers for data are Front office (Trading/Portfolio Management), Performance & Attribution, Risk & Compliance & Back & Middle Offices. In most cases, the functions have their own specialized applications that consume and interpret data in its own unique manner in addition to the unique Business Requirements of these teams.

The disparity has further been complicated by the

• source of the data (vendor and /or data sets from same vendor) across
• unique applications used by these functions

Interestingly most of the applications used across the organizations are inter-connected. The inter-connected applications push the expectation bar up on consistency of data across applications.

The EDM:

Enterprise Data Management Systems are expected to be a single point of data source within the organization that can somehow magically resolve all the data inconsistency issues across various departments within the organization, optimize the data costs (with the fond hope to reduce the data costs) and have a repository of all data to avoid future inconsistencies. Well do the EDMs in the market do this?

Answer is Yes and No.

Yes because they do help to optimize the data cost strive to become a data warehouse. Thankfully the cost of the EDM itself is not classified as data cost.

At the same time a No because they don't necessarily provide a one-stop shop for all data needs within the organization.

Challenges of EDM:

How could such an intelligent system designed to be a holistic solution does not fulfil its purpose?

The EDM does its job well by itself in terms of sourcing multiple data sets, performing the required validations and storing the data for onward distribution within the organization. It faces three primary challenges:

1. Lost-in-translation: EDM has to translate or in some cases transform the data to the format usable by the other applications.

2. Data Transformations: Data from the EDM is transformed within the applications and interpreted in ways that may not yield consistent results.

3. Latency: The perceived latency from an additional application layer is a major concern in the front office and in some time-sensitive functions across the organization.

Overcoming these challenges:

The above challenges are well founded and can be navigated with careful planning during the study and implementation phases. The key points to note are

1. Creating a Data Map
2. Maintaining distinct Data sets for each vendor
3. Defining Data Quality & Golden Source rules
4. Defining Translation Table / Transformation rules
5. Avoid All-in approach
6. Ensuring Data Users Participation across the project life cycle.

Conclusion:

All EDMs in the market strive to help manage the data effectively while being as flexible as they can be. The effectiveness of the implementation is purely dependent on the Current Statement Assessment, Solution Design, Project plan and implementation.

Note: These are my personal observations and learning from implementing EDM & Data Management functions across organizations. I do note that this list may not be exhaustive but hope it to be a worthwhile read for those embarking on the ambitious journeys.