Embedding big data analytics inside banking

By: Venkatesh Vaidyanathan, VP Product Management, Infosys Finacle

More than a decade ago, before data and analytics became the flavor of the day, South African Absa Bank employed predictive
analytics to identify and secure its branches at the highest risk of armed robbery, and thereby reduced bank robberies and cash losses by almost 40%. More recently, global financial services
major UBS Group turned to a cloud-based artificial intelligence-driven analytics solution to map the behavioral patterns of an entire nation, which could potentially transform their wealth management strategy.
The point is that banks understand the value of parsing data to address different business challenges; they don’t need to be converted, they already are data believers. But it is time to act on those data convictions. Most banks still use a minuscule proportion of their data, and harnessing external information is yet to become mainstream.
Clearly, banks need to get on with it, but they also need to do more than deploy solutions at the process or application level. Big data and analytics must be embedded deep inside banking culture and structure both. There are steps to get to this analytical end-state:
#1. ADOPT AN INTEGRATED ENTERPRISE-CENTRIC
ANALYTICS STRATEGY
Traditionally, the practice of analytics involved the centralization of all data into an enterprise data warehouse or the deployment of solutions at the application-level. Neither approach took into account the needs of every business user, nor the fact that most banking data was spread across multiple transactional silos.
But analysts report that the trend is changing. Big data deployments are becoming more decentralized. Businesses are increasingly looking for platforms that accelerate deployment across the enterprise even as they democratize the analytics practice. From banks’ point of view, this approach would help address their biggest obstacle to big data adoption, namely the presence of data silos. A platform brings together multiple
data types and sources and develops analytical use cases defined by actual business user needs, enabling banks to ramp up their analytics capabilities across the enterprise, while ensuring that technology investments are perfectly aligned with the core business objectives that define the program.

So, as a step one, banks must focus on building integrated, comprehensive, and collaborative analytics platforms that take a unified view of data and deliver solutions differentiated by business objectives and user needs.

#2. ENABLE GOVERNED SELF-SERVICE
Data volumes are exploding. Business users need realtime insights rather than curated prescriptive reporting. The traditional model of analytics as the exclusive purview of the IT organization is no longer practical or productive; banks simply have to empower business users with analytic capabilities. This means that esoteric technical protocols will have to be distilled into intuitive easy-touse self-service tools facilitating business decisions. IDC estimates that visual data discovery tools—a key driver of self-service adoption—will be the enterprise norm by 2018.
Governance is an obvious concern in a decentralized analytics paradigm. Hence, this is the new mandate for the IT organization in a democratized big data enterprise. IT will now be responsible for defining and implementing governance models that enhance data security without compromising usability. So, the focus should be on defining a role and security-based governance model that is optimized for business functionality, data security, and compliance.

#3. AUTOMATE ANALYTICS
The digital age is deluging banks with new data sources like social and mobile. The Internet of Things (IoT) will swamp that deluge. This presents a serious analytical challenge of deriving insights in real-time and at scale. One analyst says analytics will have to become more advanced, pervasive, and invisible to process the exploding volumes of data into insight, and reduce the manual intervention involved. But, in reality the next evolutionary stage for analytics is automation. In other words, analytics solutions must go beyond merely prescribing resolutions to scenarios to proactively resolving situations.
In its simplest form, automation could mean using event-based alerts that initiate rather than call for action.
In the long term, when analytics is increasingly embedded in transactional applications and business processes, automation should trigger real-time actions based on a set of business rules or sequence of events.

#4. OPT FOR OPEN SOURCE
Many pre-built proprietary analytics solutions offer the advantage of easy deployment and customization. But over the long term, the cost and time for upgrading a solution to support evolving business prerogatives, can be prohibitive.
Open source technologies like Hadoop are revolutionizing the use of big data. This has even influenced the way many proprietary solutions are being packaged and priced.
For banks, the security of open source technologies has been the key concern. There have also been other adoption challenges like the lack of enterprise-ready platforms, inadequate customization, development, and administration support. But a new breed of complete solutions is emerging, pre-packaging technology enablers, accelerators and services together with leading open source components.
Open source technologies will be a strategic tool, rapidly
expanding the scope and scale of banks’ big data efforts. Step four is clearly about building partnerships that not only deliver the evolutionary potential of advanced open source analytics solutions, but also help customize and integrate them with the existing technology ecosystem.

TURNING THE WHEELS OF COMPLEXITY INTO
SIMPLICITY
The extensive scope for advanced big data analytics technologies in banking, be it for enhancing customer engagement or building competitive advantage, is well
documented. But in order to harness the true potential of
these new technologies, banks have to take a more holistic view of deployment. Current discrete experimental applications of advanced analytics should sooner rather than later lead to an enterprise-centric open source analytics platform that empowers all business users with the tools they need to get from information to insight to outcome in real-time.