Turning AI Strategy into Tangible Value in Banking
-
September 26, 2025
Downloads
Download Article
-
Across the globe, banks are racing to harness the power of Artificial Intelligence (AI) to elevate customer engagement, optimize operations and boost profitability. However, many institutions struggle to turn their AI ambition into tangible results.
This article presents a real-world success story—how one up and coming retail bank began its journey toward becoming an AI-centric organization by proving the commercial value of AI-led initiatives through a focused, high-impact pilot.
The Challenge
Operating in a highly competitive retail landscape, the bank saw an opportunity to accelerate its growth through innovation. With aspirations to evolve into an AI-centric organization, it recognized the need to grow its customer base, modernize its technology infrastructure, and further develop its product offerings—positioning itself boldly for the future.
FTI Consulting partnered with the bank to initiate a long-term AI transformation journey. We began with a full-scale diagnostic assessment covering data, technology and AI readiness, followed by building executive alignment on a three-year transformation roadmap.
To complement the broader transformation efforts, a value creation program was launched to start generating impact early. The key challenge was to prove that the estimated business value could be realized—so we designed a focused proof-of-concept (“POC”) around a high-priority use-case to test the potential with current capabilities.
The Approach
To translate the AI vision into measurable results, the bank and FTI Consulting selected Retail Customer Value Management (“CVM”) for its focused POC. The initiative targeted cross-selling credit cards and personal loans—two key growth areas for the bank.
We began by assessing the bank’s overall CVM ecosystem, including data, technology, processes and campaign readiness. Based on this, we built a use-case that improved targeting by creating a 360º customer view and developing AI models to predict cross-sell potential. We then tested the full execution journey through a pilot campaign across multiple channels, including tele-sales and with relationship managers.
The Solution
- AI Value Creation Program
From an initial pool of over +800 AI use-cases, we prioritized 30 high-impact initiatives across CVM, Customer Experience, Digital Marketing, Bad Debt, and Fraud, among others. Given the bank's ambition to scale rapidly, the POC focused on cross-selling two products: a credit card and a personal loan. - Proof of Concept
Given existing limitations, the POC was designed around three core objectives: (1) establishing a unified 360º customer view as the single source of truth, (2) developing AI-driven models to predict cross-sell potential per customer, and (3) delivering intelligent, hyper-personalized outbound communications with adaptive scripting, real-time feedback tracking, and performance monitoring. - Pilot Execution
A multi-channel communication plan was used to drive customer engagement and conversion: (1) the customer journey began with personalized SMS messages to drive initial awareness, (2) followed by informative emails to deepen understanding. (3) Subsequently, trained agents proactively called selected customers through four coordinated channels over a three-week period. A multi-attempt contact plan ensured consistent follow-up, with up to three outreach attempts per customer before categorizing them as unreachable.
The Impact
- Observed Impact of Pilot
Since the bank’s sales process can be lengthy to turn a lead into conversion, we allowed a 4-week cooling-off period after the campaign execution to capture delayed conversions. This helped ensure accurate measurement of results. The analysis showed a strong double-digit revenue uplift from the Target Group compared to the Control, proving the impact of AI-driven targeting. - Potential of Full Program Scale-up
Based on these verified outcomes, we created a conservative extrapolation scenario, factoring in campaign eligibility, who would opt in, who would be reached and channel capacity. Even with these practical assumptions, the broader AI program is expected to outperform initial expectations—especially as structural improvements in the Bank’s CVM ecosystem will make it easier to scale.
Key Learnings
AI transformation is vital—but monetizing it is even more critical. This engagement underscores some essential lessons for banks looking to extract value from AI:
- Data is the bedrock—completeness, accuracy, and actionability at the customer level are non-negotiable: Lifestyle-related card spending patterns, fund transfer destinations and digital banking behaviors are key data points for delivering personalized customer engagement.
- AI can solve complex problems through both classical machine learning and emerging generative AI approaches: We deployed two binary classification models to predict cross-sell propensity at the customer level, resulting in a significant enhancement of targeting capabilities—achieving up to twice the conversion rate.
- True value is unlocked through cross-functional execution—aligning tech, business, and leadership: Direct engagement from leadership enabled us to dedicate over 50 sales representatives (over 5 different channels) to prioritize outreach to more than 8,000 pilot customers, supported by daily stand-up meetings to continuously refine the agile sales process.
- Hyper-personalization drives higher conversion—customers respond better to offers tailored to their unique needs: We gathered qualitative feedback from each customer contacted, with 23% of not interested customers indicating that personalized pricing and tailored content would increase their likelihood of completing the purchase.
- Channel readiness—including agent’s technology set-up, training and commission frameworks—is crucial to operationalize recommendations: Due to channel capacity and capability constraints, only 60% of identified pilot customers were successfully reached. Multiple improvement opportunities exist to increase this rate, including enhancements in technology infrastructure, agent training, and commission structures.
- Agile and smooth back-end processes are required to support dynamic campaigns - avoid delays or strict rejections in credit risk approval process: A high rejection rate impacted the pilot’s overall success, with nearly 40% of initiated card applications declined during the risk review process. Additionally, extended review times—often exceeding 4–5 working days—led to lost interest among initially convinced customers.
- Tracking each step of the customer interaction, analyzing feedback and driving contextual insights helps improve campaign success: We encountered structural challenges in tracking application status using historical logs, as the bank’s data ecosystem was designed to overwrite the status with each update. This created significant inefficiencies in monitoring the daily progress of both control and target groups.
This case proves that with the right structure and pragmatic execution approach, AI value creation is not just a promise—it’s an achievable reality for banks at any stage of maturity.
Related Insights
Related Information
Published
September 26, 2025
Key Contacts