Transaction Filtering

Transaction filtering is the real-time screening of payment messages against sanctions and watchlists before a payment is released. It is routinely confused with transaction monitoring, and the distinction is sharp: filtering runs before the payment, blocks it, and asks whether a party is prohibited. Monitoring runs after, generates an alert, and asks whether a pattern is suspicious.

Also called Sanctions filtering, payment screening, real-time screening
When it runs Before the payment is released — in the payment path
What it screens Names, addresses, countries, banks and free-text fields in the payment message
Screened against OFAC, UN, EU, UK and other applicable sanctions lists
Outcome of a hit The payment is held or blocked pending review
Liability standard (U.S.) Strict — no intent required
Regulatory example New York’s Part 504 requires both a filtering program and a monitoring program
Principal operational cost False positives, which must be cleared while the payment waits

Filtering and monitoring are not the same control

Transaction filtering Transaction monitoring
Timing Real time, before release After the fact, in batch or near real time
Question asked Is any party to this payment prohibited? Does this activity suggest financial crime?
Compared against Sanctions and watchlists The customer’s expected behavior
Effect of a hit The payment stops An alert is raised; the payment has completed
Failure consequence A sanctions violation — strict liability A missed suspicious activity report
Tuning pressure Speed against thoroughness Alert volume against coverage

Because they answer different questions, neither substitutes for the other, and regulators have said so explicitly — New York’s Part 504 regulation requires covered institutions to maintain both, and to certify the adequacy of each.

Why it is hard in the payment path

Filtering has a constraint monitoring does not: it sits in the way. Every millisecond it takes is latency on a live payment, and every false positive is a held transaction and an irritated customer.

The matching problem is the one described under sanctions lists — transliteration, aliases, entries with no date of birth to disambiguate against — but with a stopwatch running. Payment messages make it worse. Free-text fields carry names in inconsistent formats, remittance information contains arbitrary strings, and a legitimate reference can match a listed name by coincidence. A payment mentioning a city that shares a name with a designated vessel will hit.

The tuning choice is therefore harsher than in monitoring. A false negative in filtering is a strict-liability sanctions violation; a false positive is a delayed payment. Institutions tune toward over-matching and absorb the review cost, which is why filtering programs are judged on how quickly hits are cleared as much as on what they catch.

Why this matters for identity verification

Filtering screens whatever the payment message contains. If the name in that message came from an account opened without the identity being properly established, the filter is screening a string rather than a person — and doing so accurately, which is the trap.

Two consequences follow. An account opened under a synthetic identity is filtered under a name that belongs to nobody, so every payment clears legitimately and the control has been satisfied without being exercised. And weak identity data makes hits unresolvable: matching a common name against a list entry with no distinguishing detail cannot be cleared quickly when the institution itself holds no date of birth or document number to compare.

Identity verified against an authenticated document supplies exactly what filtering needs to work at speed — a name as the issuing authority spells it, with corroborating detail to disambiguate against. That is the practical argument for treating identity document verification as an input to AML, PEP and sanctions screening rather than a separate obligation, and it is where most of the false-positive burden can actually be reduced.

What transaction filtering can’t do

It does not detect suspicious behavior. A payment between parties on no list passes, however odd the pattern. That is monitoring’s job.

It cannot see beyond the message. Filtering reads what the payment carries. Ownership structures that make a counterparty blocked under the 50 Percent Rule are not in the message.

It does not validate the identity behind a name. The string is screened, not the person.

Clean filtering is not a defense. Sanctions liability is strict; screening diligently and missing a party remains a violation.

Frequently asked questions

What is the difference between transaction filtering and transaction monitoring?

Filtering screens payments in real time against sanctions and watchlists before release, and a hit stops the payment. Monitoring reviews activity after the fact against the customer’s expected behavior, and a hit raises an alert for investigation. They answer different questions and regulators generally expect both.

What does transaction filtering screen against?

Applicable sanctions and watchlists — OFAC’s lists in the United States, plus UN, EU, UK and other regimes depending on the institution’s exposure. It screens names, addresses, countries, intermediary banks and free-text fields carried in the payment message.

Why do filtering systems produce so many false positives?

Because names vary in transliteration and spelling, many list entries carry no date of birth or identification number to disambiguate against, and payment messages contain free text that can coincidentally match a listed name. Institutions tune toward over-matching because a missed match is a strict-liability violation.

Is transaction filtering legally required?

Sanctions compliance is mandatory, and filtering is the standard means of achieving it for payments. Some regulators are explicit about the program itself — New York’s Part 504 requires covered institutions to maintain both a filtering program and a monitoring program and to certify each.

Related reading

Discover Our Solutions

Exploring our solutions is just a click away. Try our products or have a chat with one of our experts to delve deeper into what we offer.

Report
Mapping the Rise of AI-Powered Identity Fraud

AI didn't just make fraud faster. It made it a system. We analyzed millions of identity interactions to map how identity attacks are evolving across regions, attack types, and sophistication levels — and what organizations need to rethink to keep pace.

See the Data