THE SMART MONEY: Decoupling financial services for cross-departmental big data integration

Data modeling

The foundation for holistic customer views is the ability to de-compartmentalize various business units and their myriad databases for singular integrations across these points of variation. That foundation similarly applies to advantageous trading, fintech capabilities, and options for profiting from open data. Although there are numerous approaches for achieving this end, Deloitte has recently recognized the value of paving such integrations with the standardized models characteristic of knowledge graphs. These models evolve to incorporate additional business requirements or sources, simplifying the schema concerns that impede many integration efforts. Organizations can thus implement a common model of all the things that are important to their company, independent of what they are called in any database, Aasman explained. The result is a range of heightened capabilities, including the following:

Exchanging information: Common models not only enable information sharing among business units, but also between external, alternative sources and internal ones.

Mapping: Modeling standards help financiers look across all of source data systems and figure out that a customer in one system is a customer in another system, and how they connect to the model, Loughlin observed. This functionality is critical for entity resolution, not only for individual customers but also for companies.

Standardizing concepts with taxonomies: Pinpointing the specific concepts in holistic models enhances the ability to describe them with taxonomies that preserve the meaning of data across different systems, simplifying integration attempts.

Data cataloging concepts

When decoupling data across sources for singular use cases such as Customer 360 initiatives, effective data cataloging is requisite for storing, understanding, and capitalizing on the many business concepts contained in common data models since, Aasman noted, many organizations “have a thousand databases with overlapping concepts.” There’s a virtuous cycle between data catalogs and common business models. The former provide a locus to inventory all of your data sources so you can register them and pull in the technical data and make models from it, Loughlin said. “But beyond that, you can start applying concepts to that metadata—the business concepts.” Those concepts make business models more expressive and specific for enterprisewide use cases.

Accurate cataloging is pivotal for reusing data assets across departments and databases. It involves comprehensively looking at all databases as assets and describing all the content of the data assets in the catalog, Aasman pointed out. Subsequently, organizations can leverage those descriptions in the catalog to determine relevant information for credit card discharges for a specific customer, for example, to determine relevant mortgage or loan information. The action derived from data cataloging is the basis for leveraging open data or trading opportunities in finance, too. According to Loughlin, contemporary cataloging choices don’t just denote where germane data exists, but are necessary to “link it to business concepts and then actually operate on the data itself.” For example, a company might want to integrate internal and external data sources to see which telecommunications package could best accommodate a certain customer’s lifestyle.

Decoupling the enterprise

The capabilities of unbundling the bank are part of a larger, ongoing development in which organizations fluidly access and act on data for cross-departmental use that frequently necessitates external, distributed sources. Rapid, ad hoc big data integrations are necessary to increase customer satisfaction. Capitalizing on trading, open data, Customer 360 initiatives, and fintech opportunities are predicated on de-compartmentalizing what were once monolithic processes—and balancing any risk associated with doing so. “You’ve got to have good risk management practices, but that doesn’t curtail your ability to monetize data,” Kwiatkowski concluded.