The Screening Decision Record
A screening decision record should distinguish the raw result, the coverage actually executed, the candidate set considered, the human disposition, and the retained evidence state. A match score is none of these.
An open reference, not an industry standard, not legal advice.
A score says what. A record says why.
A percentage cannot tell an auditor which alias matched, which source was current, what counted against the candidate, whether the candidate set was complete, or whether an unavailable list was mistaken for a pass. The record below is the minimum set of facts needed to reconstruct a screening result and to judge whether a no-hit was eligible to be treated as clear.
What the record contains
input.raw + input.normalized_query
The subject as submitted and the query actually screened. Distinguishing the two shows WHAT was screened after normalization/transliteration.
screening_mode
standard (recall-oriented, can support clearance) vs exact_only (a narrow lookup that never earns a compliance clearance).
sources[]
Every source consulted, with its exact list_version and state (current / stale / unavailable / disabled / plan_excluded) at screening time.
coverage.status
complete / partial / unavailable. Partial or unavailable coverage cannot support a clearance, an unanswered in-scope source is on the record, not hidden.
candidate_set.complete
Whether the candidate set was exhaustive or capped/truncated. A capped set cannot clear.
reasoning[]
The signals that moved the result (supporting/conflicting). Retrieval is candidate retrieval, not an analyst-confirmed identity.
served_decision
clearance_eligible / review_required / monitor / block / unavailable. PEP exposure or weak resemblance must not be rendered as a sanctions-grade block.
clearance_eligible
Eligible only if mode is standard, coverage is complete, the candidate set is complete, no actionable match remains, and evidence is durable. The legal decision stays the customer's.
human_disposition
The analyst action, if any, recorded separately from the machine result. A machine result and a human disposition are different facts.
evidence.state
available / pending / unavailable, whether the decision can be reconstructed later.
policy_context + timestamps
The policy version and plan scope under which the decision was made, and when it was screened/recorded.
Full field definitions, types and enums are in the machine-readable JSON Schema. The annotated example shows the coherent case where a no-hit is not clearable because one in-scope source was unavailable.
Where this shows up in the product
- The Evidence Capsule, the retained, retrievable form of a decision record.
- Sources & Freshness, the live source states a record references.
- The Sanctions Screening API, returns these fields on every result.
Questions
Is the Screening Decision Record an industry standard?
No. It is an open reference published by Verifex, version 0.1. It has no external adoption yet. If others use, cite, fork or build against it, it may mature, but it is not presented as a standard, a regulatory template, or a claim that any result is compliant.
Is a no-hit result enough to clear a customer?
Not on its own. A no-hit is eligible for a clearance path only if standard mode ran, coverage was complete, the candidate set was complete, and the evidence is retrievable. If an in-scope source was stale or unavailable, coverage is partial and the result is not eligible to be treated as clear.
Does this make a legal or compliance decision?
No. The record reports eligibility and evidence; the legal and policy decision remains the customer's. Verifex supports compliance workflows and does not provide legal advice.