Data Warehouse Journey Evolves at Allied Irish Bank

LONDON - Allied Irish Bank (AIB) is set to increase the data volume capacity of its enterprise data warehouse after seeing benefits from improved reporting and analytics capabilities from the evolving implementation, according to Michael McMorrow, enterprise data warehouse designer at AIB Group, speaking at the Third Annual Reference Data Management for the Banking Industry conference in London in February.

The bank, which started the data warehouse journey 12 years ago, has implemented a Teradata enterprise data warehouse and developed a common data model, which is being deployed across the firm. The implementation has involved forcing the use of certain key data fields to ensure the data model is the same in all divisions. McMorrow said this design has resulted in "a lot of success with reporting."

AIB is using the warehouse approach to do financial, regulatory and Basel reporting, and analysis of customer relationship management, money laundering and credit risk.

The data warehouse, which now holds five terabytes of business data and has 500 users, includes a semantic layer ensuring all data is consistent and standardized. The platform, which is connected to source systems through an interface, feeds reporting applications, management information systems and analysts.

The source systems are responsible for providing quality data, and the warehouse may validate and refuse to accept data if it breaches a validation check. The aim is to avoid changing data in the warehouse, ensuring the data is in raw format and can be tracked.

"If you don't aim towards this sort of work (an enterprise data warehouse) you'll constantly be reacting to your reporting requirements," he said, explaining that it is crucial to build the model in a way that it reflects a bank in general - not a specific institution.

The program at AIB started with customer data in the Ireland division, and has evolved from there. "Each step has to have a genuine business case," he said, adding that the program has subsequently extended to cover risk and finance data. The implementation means that the data warehouse now provides a joined-up, single view of a customer, account and activity data for reporting and analysis by all business units.

The initial CRM project was followed by other projects, such as Basel and anti-money laundering, which leveraged off the customer data acquisition that was already in place from the CRM project, thereby substantially minimizing the data acquisition costs of those latter projects.

Meanwhile, an audience poll at the conference showed that the majority of firms do not operate with an enterprise data warehouse model, similar to the one AIB has been building.

Most conference-goers said firms operated with a data warehouse based on a loose federation of independent data marts, while some had a jigsaw-piece design, where different parts of the data warehouse had been built to fit together. McMorrow said the poll result was not surprising, but "disappointing."

To embark on an enterprise data warehouse journey, McMorrow advised firms to define their destination, and try to ensure there are benefits, so people actually want to be part of the project. He also said the focus should be on the data. "The one thing that makes us different (as a bank) is the data that we have," McMorrow said, adding that it all has to link together.

To reach the destination, a number of data issues must be considered. McMorrow said firms should question data processes and quality: How much data do you choose to capture? How good is the data you capture? Is validation built in at point of data capture? How consistent is the meaning of data across your business?