The most promising data & AI cases in Banking

For banks, ensuring the adoption of AI technologies across the enterprise is no longer a choice, but a strategic imperative. Envisioning and building the bank’s AI capabilities is critical to succeed in an increasingly competitive marketplace. The value generation potential by AI for banks is estimated to be USD 1 trillion globally per annum. The largest part of this value is to be found within the Marketing & Sales (e.g. product recommendations, client churn prevention) as well as Risk functions (e.g. fraud prevention).

Sales

Sales Force steering (e.g. “next best action”, lead scoring): Predictive and prescriptive machine-learning algorithms can help organisations find patterns in the way customers respond to different touchpoints, hence determining which actions are more likely to lead to conversion. This set of techniques is referred as “Next Best Action” (NBA) and will increase the effectiveness of your salesforce.

Customer churn prediction and prevention: It is typically multiple-times more expensive to acquire new clients than to retain existing ones. Therefore, client churn prevention needs to be a core priority for any bank. AI allows you to leverage the breadth of available client data to set-up an early warning system to identify clients at risk of leaving or withdrawing assets. Combined with a client-segment specific retention approach, this can reduce your churn rate by up to 40%.

Dynamic pricing and discount recommendation: Financial products are by definition complex, and their prices tend to be set within the context of the client relationship, with very low transparency. Therefore, predictive analytics and granular customer segmentation can be leveraged to derive a client’s product-acceptance probabilities, price sensitivities, and lifetime values. Gaining a deep understanding of a customer’s product needs, and pricing them accordingly will thus maximize price performance while minimizing customer attrition and volume loss.

Marketing

Hyper-personalized content engagement (e.g. newsletter): Achieving a real personalized banking experience demonstrates that you understand why your customers do what they do, by delivering the right customer experience, at the right time, and through the right channel. ML algorithms can enhance a more precise segmentation, delivering personal offers, and creating targeted email campaigns so that customers engage more and make transactions whenever, wherever, and on whatever device. This will result in a faster time to market, improved response rates, and better return on marketing spend.

Customised product or service recommendations: Analysing clients’ past behaviour and the integration of third-party data to a consolidated client profile allows building a recommendation engine of highly targeted personalized offerings. This mean that clients only receive notifications and marketing collateral for services they are most interest in. This results in more engaged clients and an increased conversion rate.

Alternative data acquisition (e.g. target clients based on “life events”): Investigating alternative data sources, either directly or via third-party data providers, can provide you additional insights into a customer’s habits and upcoming life events. Feeding your AI models with such alternative data together with fine-grained transaction data allows you to build targeted offers to your customers. For instance, come up with a personalized mortgage plan when they plan to buy a house, or propose an investment solution, if they are about to experience a liquidity event (e.g. pay-out of pension fund due to retirement).

Back-office / central functions

Onboarding/ KYC: Customer onboarding can be lengthy and costly due to increasingly stricter KYC and AML requirements, associated with inefficient processes. Advanced data analytics help mitigate your money laundering and financial crime risk by automatically identifying PEPs (politically exposed persons) or flagging potential criminals via negative news monitoring.

Fraud detection and prevention: Fraud and cyber-crime are a massive and growing problem for financial institutions which lose up to 5% of their revenue to fraud every year. Yet, Machine Learning & AI technologies can be leveraged to detect suspicious transactions and protect your organization from fraudsters. This can be done, for instance by screening for anomalous behaviour and proactively raising alerts in case of suspected market manipulation.

Customer support (e.g. chat bots, voice bots): Using chatbots and voice bots trained on customer interaction data from previous text or voice exchanges can help in reducing the workload of human customer support for standard queries (e.g. “What’s my account balance?”). Additionally, as the bots are available 24/7, you can benefit from an increased customer service level and lower labor costs.

Credit decision making: Modeling customer solvency is the critical step that precedes any credit decision. AI-driven credit scoring lets you identify optimal lenders, customize repayment plans based on their history with your brand, and pre-approve new candidates in seconds, leading to more loan approvals and attracting borrowers who were previously overlooked.

Smart process automation: RPA (robotic process automation) becomes even more powerful when combined with AI. For example, it allows banks to quickly and accurately extract data from scanned documentation and decide on and initiate the required next process steps (e.g. process a credit card request).

Data platforms & analytics

Build-up of data platforms/ data lakes: A very common obstacle to achieving full effectiveness in the use of advanced analytics is a siloed organizational structure across multiple systems. Designing, implementing, and maintaining a central data platform that takes into account your long-term goals is key to a successful, long-lasting AI strategy.

Data analytics and insights visualization for decision-making: Instead of using different dashboards to track a single client, advanced data analytics and visualization tools provide personalized portfolio aggregation and reporting for investment providers, delivering insights using multiple data sources to serve clients in a more personalized manner. Single client view platform and entity resolution

Integrated understanding: Clients with multiple accounts in your subsidiaries might hold different risk positions across locations. It is then crucial for you to have a unified look at a client’s holdings, identity, and risk across siloed systems. The resulting gain in visibility on global customer activity will allow you to assess one’s risk exposure to money laundering and other high-risk financial activities. Hence, machine-learning capabilities can significantly accelerate the entity resolution process by automatically surfacing likely matches and high-risk client profiles (e.g. customers who have been issued too much credit).

Asset management/ trading

Rogue trading prevention: Rogue trading, or fraudulent trades conducted by bank employees, can cost organizations billions of dollars. A key challenge in systematically reducing this risk is that the involved data sets are too large for manual assessment. However, AI and machine learning technologies can detect anomalies based on traders’ daily behaviour, resulting in the prevention of operational and regulatory losses, brand damage, and substantial regulatory fines.

Data analytics to identify outperforming strategies (e.g. sentiment analysis): AI-driven research and modern data analytics show potential that new trading strategies that are complementary to traditional models can be discovered. For instance, testing whether buying stocks on media sentiment might lead to outperformance is now feasible thanks to the increasing computing power of cloud computing and the complex natural language processing of online news and social media.

Knowledge platforms (e.g. “What is our position on …?”): Modern knowledge platforms support data integrated from multiple sources. These platforms in turn create the possibility, in near-real time, to offer ultra-personalized experiences and run advanced analyses for decision-making based on natural language text queries (e.g., “What is our position on…?”).