# Lloyds Banking Group Achieves High Accuracy in Property Identification with AI

> Discover how Lloyds Banking Group harnessed the power of AI and UPRNs to revolutionize address matching, tackling persistent data challenges in the financial sector and enhancing customer experiences.

**Source**: ordnancesurvey.co.uk | **Published**: 2026-10-14 | **Type**: case_study

## Key Facts

- LBG's PropertyLM-UK achieved 94.6% accuracy, showcasing AI's potential in data management efficiency.
- The system's 99.64% precision in confident matches highlights a significant competitive advantage in accuracy.
- Avoiding incorrect matches in 92.6% of absent properties reveals a strong risk management capability.
- Fast processing time of 352 ms indicates operational efficiency, enhancing customer experience and service speed.
- UPRN integration supports greener financing, aligning LBG with sustainability trends and market demands.

## Summary

\## Summary
Lloyds Banking Group (LBG) faced challenges in accurately matching various representations of UK addresses across multiple systems. To address this, they developed PropertyLM-UK, an AI-based address resolution system that significantly improved matching accuracy. The system achieved a 94.6% success rate in identifying correct property references, enhancing operational efficiency and data integrity.

\## Background
Lloyds Banking Group is a major financial services organization in the UK. Prior to the deployment of PropertyLM-UK, LBG struggled with inconsistencies in property address data, which was recorded in various formats across customer applications, internal systems, and external datasets. This complexity hindered their ability to manage property-related information effectively.

\## Challenge
The primary challenge was the difficulty in matching different versions of the same UK address. Traditional methods often failed due to variations in spelling, formatting, and missing information, leading to inaccuracies in property identification. LBG sought to create a more reliable and automated address matching tool that could handle these discrepancies while maintaining transparency in decision-making.

\## Solution
LBG developed PropertyLM-UK, an innovative address resolution system that integrated AI with deterministic checks. The system utilized the OS Open Unique Property Reference Number (UPRN) dataset, AddressBase Premium for authoritative addresses, and Energy Performance Certificate (EPC) data. PropertyLM-UK connected various address representations to a stable UPRN, allowing for consistent property identification. The AI component processed over 18.2 million property records to identify potential matches, while deterministic rules ensured that critical address components were verified before accepting a match.

\## Results
PropertyLM-UK achieved a 94.6% success rate in identifying the correct UPRN as the first result from a set of 10,000 test queries. The top five results contained the correct UPRN in 97.1% of cases. When the system was confident in a match, it reached a precision of 99.64%. For properties absent from the reference index, it correctly avoided making a match in 92.6% of cases. Additionally, the system processed queries in a median time of 352 milliseconds on standard CPU hardware.

\## Key Insights
The deployment of PropertyLM-UK highlights the importance of combining AI with deterministic rules to enhance data accuracy. The 'human in the loop' approach ensures that automated systems do not operate in isolation, allowing for human oversight when necessary. This model can be applied to other sectors facing similar data challenges.

\## Customer Testimonial
No direct quote is provided in the source material.

## Entities

- **Companies**: Lloyds Banking Group
- **Products**: PropertyLM-UK
- **Technologies**: AI, UPRN, Energy Performance Certificate (EPC)
- **Organizations**: OS

## Key Concepts

Address matching, AI-based tools, Unique Property Reference Number (UPRN), Human in the loop, Data transparency, Deterministic checks, Property identification, Energy efficiency

## Definitions

- **Address matching**: The process of identifying and linking different representations of the same property address across various datasets.
- **Unique Property Reference Number (UPRN)**: A unique identifier assigned to every property in the UK, used to facilitate accurate address matching.
- **Human in the loop**: An approach that incorporates human intervention in automated processes to ensure accuracy and accountability.
- **Energy Performance Certificate (EPC)**: A document that provides information about a property's energy efficiency and environmental impact.
- **Deterministic checks**: Strict rules applied to critical components of an address to ensure accurate matching and prevent incorrect associations.

## Use Cases

- Connecting address records with Energy Performance Certificate information
- Identifying properties for energy-efficiency improvements
- Supporting greener financing propositions
- Customer onboarding in rental and commercial property markets
- Distinguishing between multiple units within one building
- Understanding complex premises with multiple owners or occupiers

## Frequently Asked Questions

**What is PropertyLM-UK?**

PropertyLM-UK is an AI-driven address resolution system developed by Lloyds Banking Group that combines AI with deterministic checks to accurately match property addresses.

**How does the system ensure accuracy in address matching?**

The system uses AI to understand variations in address formats and applies strict checks on critical address components, ensuring that only valid matches are accepted.

**What role does the Unique Property Reference Number (UPRN) play?**

The UPRN serves as a consistent identifier that links different representations of an address to the same physical property, enhancing the accuracy of address matching.

**How does the 'human in the loop' approach work?**

This approach allows for human review of ambiguous cases, ensuring that decisions made by the AI system are supported by evidence and can be audited for transparency.

**What are the benefits of using PropertyLM-UK?**

PropertyLM-UK provides a reliable foundation for property identification, improves efficiency in address matching, and supports initiatives like energy efficiency improvements and greener financing.

## Links

- [Read on Welcome.AI](https://welcome.ai/content/lloyds-banking-group-achieves-high-accuracy-in-property-identification-with-ai)
- [Original source](https://www.ordnancesurvey.co.uk/customers/case-studies/lloyds-banking-group)

---

Source: Welcome.AI | https://welcome.ai/content/lloyds-banking-group-achieves-high-accuracy-in-property-identification-with-ai