DATA & DATA ANALYTICS
BANKING’ S DATA QUALITY CHALLENGE
At Lloyds Banking Group, data governance determines whether AI applications succeed or fail. Amit Thawani, Chief Information Officer, Lloyds Banking Group – who oversees insurance, pensions and investment divisions – faces the challenge of managing 250 years of customer data spread across legacy systems.
Amit puts it bluntly: poor quality data leads to unreliable models and potentially disastrous outcomes.
This reality drives Lloyds’ approach to data infrastructure, where consolidation comes before innovation. The bank is“ heavily investing on first getting our data right,” Amit says, before going about implementing AI solutions.
The scale is daunting.“ We have a massive amount of data – but is this
data any useful?” Amit asks.“ Is it like a single point of data, single product data or holistic data?”
Legacy banking systems often trap valuable information in silos, making comprehensive customer views difficult to achieve.
Beyond consolidation, Amit’ s team must establish data lineage and maintain audit trails for regulatory compliance. When AI models make decisions affecting customers, the bank needs complete transparency about the underlying data and logic.
“ Since if you make some decisions based on some models, and if somebody has challenged our decision, we should be able to prove what was behind this model,” Amit explains. equation, and the human element of governance where the real work begins. As Maria emphasises, a successful data governance programme distributes responsibility away from a central IT function and embeds it within the business itself, establishing a network of people who understand and are accountable for the data in their domain.
From a digital swamp to curated library Without a human-led approach to governance, the corporate data landscape can become fragmented and inaccessible, with pockets of valuable information trapped in siloed systems. Different departments can also use conflicting metrics, leading those in the boardroom away from decisive action.
Analysts and data scientists, who should be uncovering breakthrough insights, spend up to 80 % of their time simply trying to find and clean the data they need, a CrowdFlower Data Science Report finds. Data scientists spend about 60 % of their time cleaning and organising data and 19 % collecting data sets – showing that, without effective data governance,
224 September 2025