ISO 20022: this time, let's get payments data right

Parth Desai, Founder and Chairman, ioNova AI, explains how structured addressing can turn ISO 20022 from a deadline into a strategic advantage.

 

 

The pattern we keep repeating

The ISO 20022 migration is the payments industry's best chance in thirty years to fix cross-border payments data quality – permanently. But we are at risk of squandering it.

Over three decades in financial services technology, I have watched a pattern repeat. A major migration is announced. Institutions invest in format compliance – new schemas, upgraded interfaces, connectivity tests. And the harder question – whether the underlying data is actually ready – gets deferred until it becomes a crisis.

• I watched it with SWIFT MT formats in the 1990s.
• Again, with SEPA migration, where IBAN adoption was the finish line, while address data remained inconsistent.
• And with real-time payment schemes, where go-live celebrations preceded the discovery that data could not sustain straight-through processing.

Each time, institutions that distinguished between technical compliance and genuine data readiness emerged stronger. We are in this cycle again. But this time, I believe we can break the pattern.

This time is different – and here's why

Three things have converged that have never aligned before.

The mandate is coordinated and unambiguous. The European Payments Council has ruled that after 15 November 2026, unstructured addresses will no longer be allowed in SEPA payments. SWIFT's CBPR+ programme has made a parallel commitment – only structured or hybrid formats will pass. The BIS/CPMI has set end-2027 as the target for fully operationalising harmonised data requirements under its G20 roadmap. There is no jurisdiction to hide in.

The technology to transform unstructured data into fully structured, compliant output at enterprise scale is production proven. Purpose-built platforms combining deterministic processing, context-aware NLP, and deep payments domain knowledge are processing billions of transactions annually across major global institutions, at sub-50-millisecond latency with complete audit trails.

And the momentum is real. Over 3.9 million CBPR+ messages flow daily on the new rails. The industry is already moving. The question is whether we use this momentum fully – or settle for ‘good enough’ again.

The prize is far bigger than compliance

Compliance is the floor, not the ceiling. The BIS/CPMI has noted that six out of ten cross-border B2B payments require manual intervention. When the full cost is aggregated -payment repairs and investigations, false-positive sanctions investigations, correspondent bank queries, and delayed payments - the total annual burden runs to an estimated $15–25 billion, absorbed invisibly across operations, compliance, and customer experience.

Fully structured payments data changes this calculus. Institutions deploying production-grade address resolution report straight-through processing rates exceeding 98%, with the joint EY/SWIFT ISO 20022 study and Citi TTS research estimating that structured party data reduces screening false positives by 25–30%. Exception costs drop from $25 per case to under $5.

The value extends beyond operations. Structured addresses enable entity resolution across KYC systems, geo-risk scoring, and fraud detection — and feed clean inputs to AI tools, where industry research consistently cites data quality as the leading obstacle to success. Structured addressing also supports FATF Recommendation 16 and EU Regulation 2023/1113 on wire transfer obligations. This is not a regulatory burden. It is an infrastructure upgrade with compound returns.

The barriers are now removable

The obstacles are well known: legacy systems storing addresses as free-text blobs, corporate ERPs passing single-line fields into payment instructions, and hybrid addresses treated as permanent rather than transitional, despite PMPG guidance documenting recurring violations in format, field population, and placeholder values.

But every barrier is now solvable.

• Regex-based parsing cannot distinguish Paris, France from Paris, Texas, but purpose-built NLP with multi-signal geographic disambiguation can.
• Manual remediation cannot meet the ISO 20022 deadline, but automated resolution processes millions of records with sub-second turnaround.
• Generic large language models cannot guarantee deterministic, auditable output, but domain-specific platforms deliver it with 100% accuracy and full explainability for every parsing decision.

The problem of converting free-format addresses into fully structured, compliant output is solved. The constraint is no longer technical capability – it is institutional decision-making speed.

Let's act together – now

No single institution can fix this alone; it spans originators, correspondents, corporates, and infrastructures. But every institution can start with what it controls.

• Validate early, not at the network boundary.
• Treat hybrid as an engineered fallback, not a default.
• Remediate customer master data as a data programme, not a message-mapping project.
• Quantify the value by connecting improvements to STP rates and screening precision.
• And front-load quality by deploying intelligent resolution at capture, during remediation, and across the payment pipeline.

Over thirty years, I have learnt to look for one signal: does the investment compound across the organisation, or serve a single compliance checkbox? Structured addressing compounds. That is what makes it strategic.

Structured by default. Hybrid as fallback. Unstructured eliminated.

The ISO 20022 deadline is approaching. Technology exists. The prize extends far beyond compliance. For the first time, we have the technology, the mandate, and the momentum to get this right.

Let's seize this opportunity – now.

About the author

Parth Desai is Founder and Chairman of ioNova AI. With over 30 years at the intersection of artificial intelligence and financial services, he has led the design and deployment of AI-driven systems across 100+ financial institutions in 55+ countries. A graduate of IIT Bombay and Georgia Tech (AI), and a research scientist under AI pioneer Roger Schank at Yale, Parth is a recognised voice on payments data quality and regulatory technology.

 

 

About the company 

 

ioNova AI is the entity intelligence company, providing AI-native address resolution technology purpose-built for financial services. Its platform transforms unstructured payment address data into fully structured, ISO 20022-compliant output across 195 countries and 50+ scripts, at sub-20-millisecond latency with complete audit trails. Deploying in 10–16 weeks via sidecar architecture with no legacy system replacement, ioNova enables banks and payment processors to achieve 98%+ straight-through processing rates ahead of the November 2026 mandate.