Predictive modelling, a key component of AI, uses complex patterns in data to inform decision-making. In banking, it enhances applications ranging from individualised mortgage pricing to credit-risk assessments and algorithmic trading systems. These models help banks achieve higher portfolio margins, lower churn and increase trading efficiency and profitability.

Optimisation models play a crucial role by identifying the most efficient solutions within various constraints, such as volumes, pricing and risk factors. For example, with secured loans, they improve pricing strategies, boosting portfolio margins. In wealth management, optimisation helps in client profiling and product distribution, ensuring services are tailored to individual client needs.

The principles of psychology and decision-making are also applied in AI-driven analytics to understand how individuals make financial choices, which aids in designing products that resonate with customers, thereby enhancing satisfaction and retention.

Furthermore, tailored AI solutions and strategic AI implementations empower banks to meet specific business needs and stay competitive. These customised models support functions such as third-party manager selection in asset management, optimising investment outcomes and enhancing processes in hard-money lending with faster loan processing and increased transaction volumes.

The success of these AI applications is evident across various banking sectors. For instance, upgraded individualised pricing models in mortgage loans lead to better financial outcomes. Payment divisions benefit from AI-driven cross-selling strategies that boost retention rates. And commercial banks see stronger B2B (business-to-business) relationships and increased loan uptake due to customised lending solutions.

Overall, the integration of AI and advanced analytics into banking not only refines customer service and operational efficiency but also provides banks with a strategic edge in a competitive market, heralding a new era of banking that is more agile, innovative and customer-focused.

Despite the potential benefits of AI and advanced analytics, many banks struggle to implement these technologies effectively. To address this, it’s important to understand the common pitfalls so that they can be avoided in future projects.

Large change projects are notoriously prone to failure, and most digital-transformation projects fail. Investments in digital-transformation projects are often wasted due to this high failure rate. Loss-making change projects also happen to be the main reason chief executive officers are fired. 

As banks increasingly focus on AI-driven analytics, their massive projects often end in costly failures. These debacles are not only complete wastes of resources but may also result in unjustified dismissals of competent professionals. 

Digital transformations, much like many organisational change projects, tend to fail for a variety of reasons that often intermingle. The most common causes include a lack of leadership, poor decision-making leading to overreach, insufficient focus and a lack of specificity in defining which specific business problems need solving and precisely how they should be addressed. This often results in the failure to articulate a clear and compelling vision in change communications, coupled with a reasonable lack of organisational conviction that the proposed grand solutions can produce transformative outcomes.

These issues are further exacerbated as banks and other organisations invest in digital projects to become increasingly data-driven, leveraging artificial intelligence and advanced analytics.

There are three critical—and paradoxical—phenomena that lead to failure, either in isolation or in combination. In the following paragraphs, I will outline these paradoxes to explain why many AI initiatives falter and offer a more pragmatic, data-driven approach to digital transformation in the banking sector.

In the modern banking landscape, large investments in digital-transformation projects are often driven by emotion, hype and fear of being left behind rather than rationales, facts and data. Currently, enormous attention is being given to projects geared towards AI-driven analytics. Many banks embark on massive AI projects that promise to revolutionise their operations but end up failing at staggering costs.

These spectacular failures not only waste vast amounts of resources but also lead to unnecessary competent professionals’ terminations. The mistakes made are reminiscent of earlier technology-investment failures in the finance industry, with doomed data-lake projects as one point in case and premature and underperforming chatbot solutions as another.

It is useful to beware of three common paradoxes relating to investments in digital-transformation projects in general and in AI and analytics projects in particular.

Firstly is the Decision Paradox. Banks often proclaim a shift towards a data-driven culture. However, their decision-making processes tell a different story. Time and again, I’ve seen banks invest in digital projects based on emotion, outside pressure, psychological manipulation or executive “gut feeling” rather than hard empirical data. This paradox is not only ironic but also costly. Without grounding decisions in data, banks venture into projects misaligned with their real needs, leading to failed initiatives that drain both time and financial resources.

Secondly is the Size Paradox. The allure of large-scale digital projects is undeniable. Vendors and consultants often promote these grandiose plans, which promise transformative outcomes and are accompanied by hefty consultancy fees. Yet, the complexity and sheer scale of these projects set them up for failure, and a large majority of them do fail. In contrast, smaller, more focused initiatives often lead to success. These projects are manageable and adaptable, allowing institutions to iterate and refine their approaches based on real-world feedback and evolving technological landscapes.

Thirdly is the Solution Paradox. In many cases, banks invest heavily in broad, supposedly versatile solutions that subsequently require significant customisation to address specific operational challenges—if they ever truly fit at all. This backward approach of finding problems to fit pre-packaged solutions leads to resource misallocation. Banks should instead focus on identifying specific issues and then seek or build tailored solutions that directly address those needs, thereby enhancing efficiency and effectiveness.

Relationships between banks, on the one side, and vendors and consulting firms, on the other, are often fraught with conflicts of interest. Driven by the potential for lucrative contracts, consultants and vendors may push for larger, more expensive projects that don’t necessarily align with a bank’s strategic interests. This misalignment not only distorts the decision-making process but also prioritises the project’s size over the bank’s long-term success. Recognising and mitigating these conflicts is crucial for banks that wish to retain control over their digital strategies and ensure they align with their core business objectives.

To navigate these complexities successfully, banks must adopt a pragmatic approach to digital transformation. This includes incremental and evidence-based decision-making and developing in-house capabilities. Instead of overhauling systems in one fell swoop, banks should adopt a step-by-step approach that allows them to test new technologies on a small scale, measure the results and then scale up successful practices.

By building machine learning (ML) solutions internally to solve small, specific business problems, banks not only reduce reliance on external vendors but also develop deeper understandings of their own data- and system-integration needs. This grassroots approach to upskilling, experimenting and building in-house capabilities helps in managing risks and tailoring solutions so that digital-transformation projects are closely aligned with the bank’s strategic goals. Only after a bank has experimented, tested and found a raft of data-analytics solutions can it estimate the need and scope for an integrated solution, such as a data lake. Through these experiences, a surprising insight could emerge: It may not even need a data lake to start leveraging AI and advanced analytics effectively.

As we look towards the future, the key to successful digital transformation lies in understanding and strategically navigating its inherent paradoxes. By focusing on data-driven decisions, right-sized projects and bespoke solutions, banks can circumvent common pitfalls. Moreover, this careful, measured approach is essential not just for avoiding failures but for fostering long-term success and stability in an increasingly digital world.

Ultimately, the banking sector stands at a crossroads where the choice of path will determine its role in the digital age. Banks that leverage AI and advanced analytics judiciously will not only survive but also thrive, transforming potential disruptions into opportunities for innovation and growth. The path forward isn’t laden with sweeping changes but with thoughtful, incremental and strategic advancements that collectively enrich banking experiences for all stakeholders involved.

LeackStat 2024