The transition to ISO 20022 represents one of the most significant transformations in modern payments infrastructure. While much of the industry’s early focus has been on message format migration, from legacy MT to richer MX standards, the reality is that compliance extends far beyond syntax. One of the most critical challenges in this shift is the handling of unstructured address data, which highlights a broader requirement: payment information must be structured, granular, and consistently interpretable across systems, institutions, and jurisdictions.
This shift elevates data quality from an operational consideration to a strategic requirement. Financial institutions are now expected to process transactions enriched with detailed information that supports compliance screening, improves transparency, and enables automation across the payment lifecycle. However, as many banks have discovered, achieving this level of data readiness is not simply a matter of upgrading messaging infrastructure. It requires addressing long-standing inconsistencies embedded in upstream data sources.
Among the most persistent of these challenges is the handling of beneficiary address data, an area where legacy practices and modern requirements often diverge.
Despite the increasing enforcement of structured data requirements by global payment networks, a significant portion of inbound payment instructions continues to originate from systems that rely on free-text input. Corporate ERP platforms, treasury systems, and manual entry channels frequently generate address data in inconsistent, multi-line, and often ambiguous formats.
For banks, this creates a structural mismatch. On one hand, ISO 20022 mandates clearly defined address components such as street name, building number, city, postal code, and country. On the other, incoming data may include incomplete fields, mixed languages, abbreviations, or formatting variations that are difficult to interpret programmatically.
A regional bank in Malaysia leveraging the IMS Payment platform encountered this challenge at scale. Unstructured address data led to repeated operational friction, including validation failures, increased exception queues, and delays in payment execution.
These issues were not isolated, they had a cascading impact across the payments lifecycle: reducing straight-through processing (STP) rates, inconsistencies in compliance screening, and ultimately affecting customer experience. What initially appeared to be a data formatting issue quickly became a broader operational constraint.
Historically, banks have relied on rule-based parsers and static data dictionaries to interpret unstructured fields. While effective in controlled scenarios, these approaches struggle to scale in environments characterized by variability and ambiguity.
Address data, by its nature, is highly contextual. The same format may represent different components depending on geography, language, or user behavior. Static rules are inherently limited in their ability to adapt to such variability, often resulting in either incorrect parsing or an over-reliance on manual intervention.
As payment volumes increase and regulatory expectations tighten, these limitations become more pronounced. The need is no longer for incremental improvements in parsing accuracy, but for a fundamentally different approach—one that combines adaptability with compliance awareness.
To address this gap, the bank implemented the AI Address Structuring capability within the IMS Payments platform. This solution is built on a combination of large language model (LLM) technology and a compliance-driven rule layer, enabling it to interpret and transform unstructured data into ISO 20022-compliant formats.
Unlike traditional tools, the model is designed to understand context. It can process multi-line, multi-language address inputs and extract standardized components such as street details, locality, city, postal code, and country. These elements are then validated and mapped into structured formats aligned with both ISO 20022 and network-specific requirements.
Crucially, the solution is adaptive. As it processes new data, it learns from patterns, corrections, and exceptions, continuously improving its accuracy. This represents a shift from static parsing to dynamic interpretation—where the system evolves alongside the data it processes.
The broader IMS platform reinforces this capability by embedding AI across the payment lifecycle, enabling higher accuracy, improved compliance, and faster processing at scale.
In payments processing, automation cannot come at the expense of control. Regulatory scrutiny, audit requirements, and operational risk considerations demand that AI-driven systems operate within well-defined governance frameworks.
The implementation of AI address structuring reflects this balance. Rather than functioning as a black-box solution, the model incorporates transparency and oversight at every stage. Low-confidence outputs can be routed through human review, ensuring that exceptions are handled appropriately. All transformations are logged, creating a clear audit trail for compliance and traceability.
In addition, feedback loops enable operations teams to refine the model over time. Corrections made during exception handling are not lost; they are fed back into the system, improving future performance. Model updates can be deployed in real time or through scheduled refresh cycles, allowing institutions to align AI behavior with internal risk policies.
This governance-led approach ensures that AI enhances operational efficiency while maintaining the rigor required in financial environments.
Following deployment, the bank observed tangible improvements across its payments operations. Manual address repair cases were significantly reduced, freeing up operational resources and lowering processing costs. Straight-through processing rates improved across both Swift and domestic payment rails, contributing to faster and more reliable transaction execution with a payment modernization.
Message consistency increased, enabling more accurate compliance screening and reducing the likelihood of downstream issues. Payment turnaround times improved, and corporate customers experienced fewer delays and rejections.
These outcomes highlight a key point: addressing data quality at the source has a multiplier effect across the entire payments lifecycle. By resolving one of the most common sources of friction, unstructured address data, the bank was able to unlock broader efficiency gains.
One of the defining advantages of an AI-driven approach is its ability to evolve. Payment ecosystems are not static; they are shaped by changing regulatory requirements, emerging payment schemes, and evolving customer behaviors.
The IMS AI model is designed to operate within this dynamic environment. As new address formats, regional variations, and compliance rules emerge, the system adapts—ensuring that institutions remain aligned with evolving standards without requiring constant manual reconfiguration.
This adaptability also supports long-term scalability. As transaction volumes grow and data complexity increases, the model continues to deliver consistent performance, enabling banks to scale operations without proportionally increasing manual effort.
The challenge of unstructured data is not unique to any single institution. Across the industry, it remains one of the primary barriers to realizing the full benefits of ISO 20022. Address data, in particular, sits at the intersection of compliance, operational efficiency, and customer experience.
What this case study demonstrates is that solving this challenge requires a shift in approach. Rather than treating data structuring as a peripheral function, it must be embedded within the core payments architecture. AI provides a practical and scalable way to achieve this, bridging the gap between legacy inputs and modern standards.
More importantly, it reframes ISO 20022 from a compliance exercise into an opportunity. Institutions that invest in data quality and intelligent automation are better positioned to achieve higher STP rates, reduce operational risk, and deliver faster, more reliable payment experiences.
As the industry moves toward a fully ISO 20022-native environment, the importance of structured, high-quality data will only continue to grow. Address intelligence is no longer optional, it is a foundational capability for modern payments processing.
By embedding adaptive AI within the payment hub, IMS Payments demonstrate how financial institutions can overcome one of the most persistent challenges in ISO adoption. The result is a more efficient, compliant, and scalable payments operation, capable of meeting today’s requirements while remaining ready for tomorrow’s demands.
Redefining transaction processing solutions
