Case Study: AI Use Case Discovery & Prioritization for a Banking Group

Banking

AI Use Case Discovery & Prioritization for a Banking Group

The Challenge: 143 Ideas, 31 Pilots and Nothing in Production

A diversified banking group had no shortage of AI ideas. Its business units had submitted 143 of them over eighteen months and 31 pilots were under way, each funded locally and judged by its own sponsor. None had reached production. The executive committee could not compare one proposal with another, and it had no view of which ideas depended on data the group did not yet have or which would be expensive to run at scale. It asked for a single, defensible answer to where the next year's AI budget should go.

Our Solution: One Scoring Method and a Funded Portfolio

Lydatum ran a ten-week program with the group's nine business units and its technology and risk functions:

• Discovery Workshops: We worked through each unit's processes with the people who run them, which consolidated the 143 submissions into 88 distinct use cases and surfaced 14 that nobody had proposed.
• Scoring Model: Each use case was scored on five criteria agreed with the executive committee: business value, feasibility, data readiness, risk and running cost.
• Evidence for Each Score: Value estimates were signed off by the finance lead of the sponsoring unit, and feasibility and data readiness were assessed by the group's own engineers.
• Portfolio & Stage Gates: The top-ranked use cases became a portfolio of 12 with named owners, and funding is released in stages as each one meets its milestones.

The same review covered the 31 running pilots, and the group stopped 19 that scored poorly or duplicated another unit's work. The scoring model and its templates were handed to the group's AI office, which now uses them for every new proposal.

The Impact: Fewer Initiatives and the First Ones in Production

The executive committee approved the portfolio in a single session, and the budget freed from the stopped pilots covered 40% of its cost.

Seven of the twelve use cases were in production nine months later, and the other five were on track against their stage gates. Business units now bring proposals to the AI office in the same format, and the quarterly review compares them on the same five criteria.

Services Delivered: AI Use Case Discovery & Prioritization, AI Roadmap & Operating Model, AI Maturity Assessment, AI Unit Economics & ROI

Technologies Used: Power BI, Snowflake, Python