The use of AI continues to proliferate among retail banks to meet demands for personalized customer experiences and enhanced fraud protection, as well as to seize internal efficiencies through automation. However, AI adoption doesn’t always deliver the expected ROI. This issue can be mitigated using several practical approaches that improve how banking providers manage AI internally.
Artificial intelligence (AI) has revolutionized most sectors today, with financial services among one of the most disrupted industries. Thanks in part to AI, retail banks and credit unions are able to engage with existing customers how and when they prefer, attract new customers with greater precision and maintain trust with advanced digital capabilities. The pandemic fueled the urgency with which banking providers adopted digital initiatives, including AI systems, to improve their agility and competitiveness. Consider that AI adoption in retail banking is expected to grow 15x between 2020 and 2030. However, adopting AI, in itself, does not guarantee success. Quite the contrary.
Since the dawn of the financial technology industry and online banking, banking providers have faced challenges adopting new technology and evolving legacy systems. AI adoption in particular is not a switch companies can simply “flip on,” and unfortunately, there is no one-size-fits-all AI plug-in to purchase. In addition, despite big investments and seemingly expert advice from vendors, many companies still make mistakes implementing AI systems.
While these mistakes are easy to make, they are serious and can result in unnecessary costs and, most importantly, a loss of customer trust. And without trust, even the most innovative solutions will have little or no impact. It is critical to recognize the pitfalls that can lead to insufficient ROI from an AI system. The following are some of the most common mistakes.
Adopting AI applications vs. AI systems: AI is designed to be implemented as part of a holistic business system, versus a patchwork of standalone features or processes. Consider the use of AI-enabled chatbots to provide customer support. This feature should be managed as an end-to-end system vs. simply an automated, surface-level application. Implementing an AI application vs. system misses its true opportunity to be integrated into a customer relationship management program or other customer experience initiative that then delivers even more value to both the customer and the bank.
AI system soup: It’s common for banking providers to adopt multiple AI-enabled processes to support multiple business functions, however, problems can occur when these processes don’t communicate with each other effectively, or when they give conflicting results and advice. This can jeopardize the customer experience and result in loss of trust if an issue occurs as a result of disparate AI engines.
Going “all-in” with an AI vendor: It’s not unusual for a company to outsource the development of their AI system as well as the management of it to a knowledgeable vendor, especially if the internal team is resource-constrained or is not confident about their AI expertise. However, this approach leads to infrastructure overhauls that are costly, slow, risky and ultimately not necessary. AI vendors will typically prioritize their own technologies or preferred partners to create a “custom” solution, which could disregard other options that are better for clients. However, the biggest issue with this approach is that the company forfeits control and autonomy over the AI systems that it comes to rely upon, which minimizes an internal team’s ability to easily and cost-effectively adapt the AI system to meet changing needs and opportunities.
The solution
The solution to these common mistakes doesn’t require spending on more technology, thankfully. Rather, there are several proven, practical approaches that retail banks of all sizes can adopt to secure more ROI from their existing AI investments.
Successful AI adoption begins with one core feature: While this may sound antithetical to avoiding AI applications, the key to a successful AI system is to first apply AI to the right feature. Banking providers should identify their most critical selling point at the core of their business model, or a “killer feature.” This feature could be the bank’s ability to offer highly personalized customer support, proactive customer alerts or advanced fraud detection capabilities. Next, one specific “minimal viable prediction” is outlined to enhance the killer feature or solve a problem related to it. An AI module using select data streams associated with that feature is applied to achieve the prediction or fine-tuned until it does. This specific approach is laser-focused on securing ROI for the core business quickly and avoids paralysis due to unnecessary data analysis.
Foster organic proliferation of AI: When this first feature is working as it should, other business applications and functions are brought online around the killer feature, progressively cleaning up and connecting data streams throughout the business to the AI system. Unlike traditional implementation approaches that can require significant data and system downtime to roll out, this incremental approach supports the organic permeation of AI. It is a fast, flexible and low-risk approach that guarantees ROI.
Above all, own AI internally: The most critical step for banking providers to secure ROI from their AI systems is to own the systems internally. A vendor will never be able to identify the killer feature like the internal team. Vendors are less familiar with the nuances of a bank’s operations, opportunities and competitors, and as a result, are limited to just keeping the lights on for an AI system; not strategically managing it to drive business efficiency and growth.
AI adoption does not have to be a burdensome initiative despite inherent complexities and risk of impacting customer trust. An incremental approach focused on the core business and managed by an internal team doesn’t require a huge outlay of resources to begin delivering ROI quickly.
