Get on Track with Your Data Analytics Evolution

What are the crucial elements for AI successes?

In my earlier blog, I spoke about how the notion of data as a strategic asset is rising in corporate banks and across the broader banking community.

While every institution’s AI journey differs, the building blocks are largely similar. As a bare minimum, implementations require having modern data infrastructure and foundational technologies such as robotics process automation and optical character recognition. But adoption depends on much more than just technology.

We've come such a long way on the technology front. My final year thesis at university involved designing a piece of software that could recognize handwriting and my task was to solve 10 numbers, from 0 to 9. I first had to convert the handwriting into grids or pixels, then I put a bunch of mathematical transformations to make it unique, then after I trained the neutral network with 300 data sets, I was able to use the software to predict the handwriting. You have to remember that this was before smartphones! Back then, most of the problems were technology-related. There wasn't enough processing power to crunch the data, or it would require a large amount of expensive hardware. It is good that technological problems are mostly behind us.

How can corporate banks then ensure that AI doesn’t just become an organizational buzzword that actually adds minimal real value? How do they articulate a clear vision for moving up the data and analytics maturity curve? In this follow up commentary, we explore the critical pillars for AI successes.

Corporate banks can move the needle on their AI adoption if they:

1) Establish data-driven leadership

The banks’ AI journey begins with leadership, with adoption requiring them to move from rules-based systems to algorithm-driven models, and to envision new forms of customer engagement [e.g., involving virtual assistants for personalized customer service, or the relationship managers (RMs) prospecting recommendations generated by machine learning models]. These require senior managements’ buy-in, and their willingness to embrace new paradigms, yet ideally not peg hard ROI targets.

Having Chief Data or Analytics Officers (CDOs or CAOs) to lend their weight and experience is invaluable to drive other stakeholders’ sponsorship; encourage a data-driven mindset; and influence the tools, technologies, thinking and culture to shape the banks’ AI journey.

2) Drive collaborative engagement teams

It is vital to have a multi-disciplinary AI team that cuts across various business, product and functional areas, and specializing in data architecture, infrastructure and governance. These individuals can ensure that a common framework to evaluate various use cases are developed, learnings are shared, and ideas cross-fertilized. As AI transforms the workplace, best practices can also be efficiently emulated, as is the case with intelligent virtual assistant learnings from retail banking that can be used to refine corporate banking applications.

Change is never easy, so these teams should be given sufficient bandwidth to experiment with data innovation with a ‘permission to fail’ mentality. This allows them to take chances to explore new AI concepts, with accumulated learnings from early failures, possibly contributing to meaningful improvements for longer-term data value creation.

3) Foster a data-centric culture

Though data analytics enhancements depend on processes and technologies, the third pillar – people, is equally critical. But building a data-driven, action-oriented dashboard with a friendly user interface for RMs is only half the challenge, convincing these staff on the credibility of the AI-based recommendations can be a greater issue. Embedding data-based decision-making requires a shift in mindset, for which the banks need to ensure collective conversations to foster a data culture and generate positive behavioral change.

While there is no exact science on what such behaviors entail, this may include nurturing the staff’s comfort level with, and trust in data analytics, and upskilling them through education and training to work toward target state behaviors. Equally critical is to encourage enterprise-wide data collaborations to reduce traditional business unit siloes. Staff recognition can be achieved through initiatives such as hackathons and datathons, with rewards given for behaviors that drive the desired data analytics cultural outcome.

4) Build up from discrete AI implementations

Determining where to begin can be daunting, and banks should start their implementations within a more narrowly defined part of a journey instead of trying to adopt overly ambitious programs.

Here, having a business-centric blueprint to lay out the vision and strategy for Big Data would align the requirements of business users against available resourcing and the implementation roadmap of IT. CDOs/CAOs should also obtain a holistic view of all key business challenges that would most benefit from better data analytics, then determine the sequence these should be tackled.

I would caution leaders from pursuing too many big AI projects with multi-year time horizons as this could risk fatigue and frustration. Conversely, while small victories build momentum and provide the proof points to demonstrate the value of AI, not having more innovative initiatives could see the bank being out-maneuvered by others pursuing programs with higher ROIs at stake. Hence both are equally valuable, and corporate banks must ensure a balance portfolio between high-profiled longer-term projects versus simpler, but highly viable quick wins.

5) Ensure strong data and model governance

In my earlier article, I opened with a quote from Elon Musk. It only seemed fitting to wrap up with another, wherein he warned that “AI could be a fundamental risk to the existence of human civilization”. His suggestion is to increase regulatory oversight on the development and implementation of AI.

However, given the nascent and evolving nature of AI technologies, and the need for a high degree of flexibility, I see regulators preferring to keep a light-touch and risk-based approach to AI. Nonetheless, while legislations are not imposing a lot of restrictions (just yet), CDOs/CAOs are still ultimately accountable for their AI governance charter, and the assignment of responsibility to other functions within their banks.

They also need to ensure the delivery of strong model governance and explainable AI to ensure ethical and compliant behavior, data privacy, transparency and robust permissioning, - all which in turn, preserves the strongest asset a bank has, that of customers’ trust. 

---

It is the dawn of an AI-enabled decision-making, and banks need to get on track with their data evolution. The saying that “data and analytics is the new oil” may be cliched, but it certainly seems to apply to the future of corporate banking.

Institutions that effectively leverage their data assets and successfully navigate AI development complexities will reap multiple benefits. These future-forward banks are the ones pulling ahead of their peers and increasing agility, elevating services and clients’ customer experiences, reshaping the competitive ecosystem and improving their financial outcomes.

The views reflected in this article are the views of the author and do not necessarily reflect the views of the global EY organization or its member firms.

With thanks to @Li-May Chew