Evals for Facctum

List ingestion and normalization

Handling of incoming sanctions/watchlist records, including enrichment and validation of screening data before it reaches the matching engine.

Update timeliness and propagation

Behavior when a list is updated — how quickly changes take effect and how already-screened populations are treated.

Impact analysis of list changes

Surfacing which customers, payments or alerts are affected when a list entry is added, amended or removed.

List coverage and provenance

Identifying which lists a decision was made against and where a matched record originated.

Name matching across variants

Transliteration, aliases, name order, spelling variation and other name-form differences in customer records.

Screening profiles and thresholds

Applying dynamic screening profiles and configured match thresholds to different customer populations.

False-positive reduction and AI auto-tuning

Behavior of tuning mechanisms that suppress non-substantive matches without dropping true hits.

Entity disambiguation

Distinguishing between distinct parties who share identifying attributes such as name or date of birth.

Inflight screening and latency behavior

Screening a payment message in real time and returning a decision within the payment flow.

Hold, release and block decisioning

Which payments are stopped, which pass, and how the outcome is recorded.

Payment message field handling

Extracting and screening the parties and fields carried in a payment instruction.

Throughput under enterprise volume

Behavior when screening high record volumes rather than isolated payments.

Transaction risk detection

Identifying activity that warrants review under AML monitoring rules or thresholds.

Monitoring rules and threshold configuration

Applying configured AML thresholds and rules to observed transaction activity.

Business entity screening (KYB)

Screening corporate counterparties and their associated parties rather than individual customers.

Complex entity and ownership structures

Handling layered or cross-border business structures during business screening.

Alert triage and prioritization

Ordering and presenting alerts so investigators work the highest-risk items first.

Match evidence and rationale

Showing why a record matched, including the matched fields and source list entry.

Disposition and escalation paths

Recording a decision on an alert and routing items that require escalation.

Alert volume and fatigue management

Behavior when queues are large, including duplicate and repeat-match handling.

Data enrichment and validation at ingest

Improving match accuracy by validating and enriching screening data before matching.

Audit trail and decision explainability

Reconstructing a past screening decision and the data state it was made against.

Data privacy and security handling

Treatment of sensitive customer identity and transaction data under global regulatory obligations.

Model validation and AI governance

Evidence supporting the behavior of AI-driven matching and tuning components.