Insights from the United Kingdom’s People with Significant Control register

6. Challenges and best practices in data analysis

This assessment revealed key challenges in analysing BO data and highlighted strategies to improve data accuracy and usability. Below are three major challenges and approaches to address these.

Challenge 1: Deduplicating persons and entities

Multiple records for the same entity or person were identified, often resulting from updates in ownership relationships (e.g. changes in shareholding) or changes in reported attributes (e.g. address updates). To resolve this, Companies House identifiers were used to identify entities, while auto-generated person identifiers from Open Ownership’s BODS-formatted data from the UK PSC register were applied to individuals. This approach enabled effective deduplication, reducing entity records by 10% and person records by 5%.

Challenge 2: Interpreting ownership interests with range-based values

The UK PSC register discloses shareholding and voting rights in broad percentage ranges (25–50%, 50–75%, 75–100%) rather than exact values. This limits granular analysis, including detecting precise distribution patterns and calculating exact percentage differences between beneficial owners of the same entity, a useful metric for assessing disproportionate control.

Compounding this challenge is the lack of structured information on direct versus indirect relationships, which constrains the ability to reconstruct full ownership chains. Without these links, it becomes difficult to differentiate between direct control and influence exercised through intermediary entities. Integrating structured shareholder data with BO disclosures would enhance the ability to trace ownership flows and better detect complex or layered structures.

Despite these limitations, this research focused on the most commonly reported ownership ranges to identify meaningful patterns. This approach allows for insights into ownership distribution and supports the detection of anomalies, even without precise percentage values or information on whether relationships are direct or indirect.

Challenge 3: Managing outdated information

The dataset includes historical entries, such as inactive or dissolved entities and ended relationships. While capturing changes over time is valuable and considered good practice, retaining outdated entries can distort current ownership structures and increase computational load. [29] Filtering out non-active entities and relationships helps ensure the analysis reflects the present state of ownership networks. In this study, removing non-active entities reduced the volume of entities by 32%, while excluding ended relationships reduced the number of ownership links by 11%.

Insights from the UK PSC register highlight the importance of reliable identifiers for accurate entity and person matching, precise ownership values for detailed interest analysis, and effective historical-data management to ensure relevance and reliability. Addressing these challenges enhances the accuracy, efficiency, and utility of BO analysis, supporting more informed policy decisions and greater transparency.