Turning 115,000 records into one version of the truth

A migration deadline and a data problem that couldn't be deferred

The bank had made the decision to migrate to a new core banking platform. What it hadn't yet resolved was the state of the data going with it. Customer information was distributed across five systems — the core banking platform, a CRM, and specialist platforms for invoice finance, asset finance, and collections — each holding a partial and often inconsistent view of the same customers. No reliable mechanism existed to connect these records. No single system knew who a customer really was.

For years, the Single Customer View had been an implicit expectation across the business — widely recognised as important, but without a formal programme, a dedicated budget, or a clear owner. The migration changed that. Proceeding without first resolving the fragmented data would do more than carry inefficiency into the new platform; it would import the risk directly, jeopardising the transformation's return on investment before it had begun.

The operational consequences of the fragmentation were already visible. Relationship Managers manually aggregated customer information from multiple systems ahead of every client meeting — a slow, error-prone process that left them with low confidence in what they were looking at. Duplicate records, anomalies, and cross-system discrepancies were common but difficult to identify or quantify. And without a consolidated view, the bank had no reliable way to assess aggregate borrowing levels, total exposure, or complex relationships across its customer base. Hidden connections — shared directorships, Ultimate Beneficial Ownership structures — remained obscured, creating blind spots in risk management that the bank couldn't see clearly enough to address.

Prove it works before committing to the full programme

The bank's leadership chose to move carefully. Rather than launching a full enterprise SCV programme, it commissioned a time-boxed Proof of Concept — a deliberate act of de-risking designed to validate both the technical approach and the business value before any larger investment was made. The PoC ran for ten to fourteen weeks, working from a one-off data batch rather than live pipelines, and was structured to deliver empirical evidence rather than theoretical projections.

The process began with the extraction of over 115,000 customer records from all five source systems — Core Banking, CRM, Invoice Finance, Asset Finance, and Collections — loaded into a dedicated, secure environment for analysis. An entity resolution engine, Senzing, then compared key attributes across all records to identify which entries referred to the same real-world customer, assigning a persistent Global Party Identifier to each unique entity. Where the bank had previously held multiple conflicting records for the same customer, it now had one.

Enrichment came next. Third-party data from Experian was layered onto the resolved records to validate entities and build out corporate hierarchies — surfacing UBO structures and directorships that internal data alone couldn't reveal. A data quality tool, Aperture Data Studio, profiled the consolidated dataset in parallel, quantifying inconsistencies and prioritising what needed remediation. The resolved, enriched data was then made immediately available to business users through dashboards and integrated directly into existing Enterprise Analytics reports — giving Relationship Managers an aggregated view of their client groups from day one of go-live, without waiting for the full migration to complete.

Hidden relationships, now visible. Migration risk, now managed.

From 115,000 fragmented records, the entity resolution process created 21,485 unique, trusted customer entities — eliminating duplicates and producing a single identifier for each real-world customer across all five systems. The enrichment and relationship-mapping process identified 18,209 candidate connected groups, including 1,293 groups linked through shared Ultimate Beneficial Owners and 2,878 through shared directorships. These were relationships the bank had been unable to see. They were now mapped, validated, and available.

For the bank's risk function, the value was immediate. Aggregate borrowing levels, total exposure, and complex ownership structures across the entire customer base became visible for the first time — replacing the manual, low-confidence aggregation that Relationship Managers had previously relied on. The integrated dashboards delivered that view directly into existing workflows, with no new systems to learn and no delay.

For the migration programme, the PoC delivered something more valuable than clean data: certainty. By establishing a proven methodology to create a reconciled, high-quality customer dataset, the project gave the migration programme a validated golden source to work from — supporting entity mapping between old and new systems and ensuring that legacy data issues would not be carried forward. The risk of corrupting the new platform with unresolved, fragmented records had been addressed before a single record was moved.

The bank enters its core banking transformation with a foundation it can rely on. The SCV that had existed for years as an implicit expectation — important but unowned — is now a validated capability, with a proven methodology ready to scale across the enterprise.

‍