How Should a Sanctions Screening Vendor Report a Check It Could Not Complete?

An honest sanctions and counterparty screen reports an unverifiable step as pending or unverifiable, never as a clean pass. When a vendor returns a green result on a check it could not actually complete, for example when it cannot resolve beneficial ownership under OFAC's 50 Percent Rule or cannot reach a consolidated list, it is manufacturing a false negative. That hidden gap transfers directly onto the compliance officer who signs the trade under FATF Recommendation 10 customer due diligence obligations.

Brent settled at $76.01, WTI at $71.41, an arb of $4.60. Whether that spread widens or flattens changes the economics of a cargo. It changes nothing about your screening burden. A counterparty that clears OFAC SDN, EU, UK/OFSI, and UN lists at $76 clears them at $60. So the question worth asking before you sign a vendor is not how fast it screens or how pretty the dashboard is. It is how the tool behaves when it hits a wall. This is a buyer's checklist, not a product review. Run the test below in your next vendor demo.

The Test: Ask How the Vendor Reports a Check It Could Not Complete

Here is the single question that separates honest screening from dangerous silence. In the demo, ask the vendor to walk you through what the system returns when it cannot complete a step. Not a hypothetical. Make them show you.

There are exactly two families of answer.

The honest answer describes a distinct state. The check did not resolve, so the system flags it as pending, unverifiable, or incomplete, and it names why. The counterparty could not be matched to a legal entity. The ownership graph terminated before reaching a natural person. A list source was unreachable at query time. In each case the tool tells you the work is not finished.

The dangerous answer is a clean pass. The vendor could not complete the check, and the system returned green anyway. When you probe, the language gets soft. "No adverse match found." "Cleared against available data." Those phrases sound like a result. They are the absence of one, dressed up as an all-clear.

A clean result and an incomplete result are not the same thing. A vendor that collapses the second into the first is generating false negatives by design.

Why a Clean Pass on an Incomplete Check Is a False Negative

A false negative in screening is a hit that should have surfaced and did not. Most compliance officers picture this as a matching failure, a fuzzy-name algorithm that missed a transliteration. That is one source. The quieter source is reporting design.

Consider OFAC's 50 Percent Rule. An entity is blocked if it is owned 50 percent or more, in the aggregate, by one or more blocked persons, even when the entity itself does not appear on the SDN list. Screening the entity name alone will return no match. That is technically true and operationally worthless. The exposure lives in the ownership chain, not the name field.

Now suppose a vendor tries to resolve that ownership chain, cannot obtain the layer of intermediate holding companies it needs, and reports the counterparty as clean. The system did not find a match because it did not finish looking. The green result is not evidence of a clean counterparty. It is evidence that the check stopped early. If that counterparty is in fact majority-owned by a blocked person through a holding structure, the clean pass is a false negative that a mandate chain or layer cake structure was built specifically to produce.

The compliance officer who signs off inherits that gap without knowing it exists. That is the transfer of risk the buyer needs to price.

What Honest Reporting Looks Like

Honest reporting keeps two categories visibly separate: what the system verified, and what it could not.

A verified clean check means the tool ran the counterparty against the lists it covers, resolved the ownership question to the depth required, and found nothing. A pending or unverifiable state means one or more steps did not complete, and the tool says which ones and why. The compliance officer then decides whether to escalate, request documents, or hold the trade.

This is the design stance OilFlow takes. The system distinguishes completed checks from steps it could not verify, and it surfaces the unverifiable steps as pending rather than folding them into a pass. The point is not that pending is a nicer word. The point is that pending preserves the analyst's ability to make a decision on accurate information. A false green removes that ability and hides the removal.

When you evaluate any vendor, insist on seeing this separation in the output itself, not in a sales deck. The report should show a pending state as a first-class result, with a reason attached.

The Second Question: List Coverage

Once you have confirmed pending-not-clean reporting, confirm what the tool actually screens against. A false negative can also come from a list the vendor never checked.

The baseline expectation is coverage of the major consolidated lists. OFAC SDN. The EU consolidated list. The UK/OFSI list. The UN Security Council list. On top of those, sectoral and export-control lists matter for oil and refined-product trades, including BIS entity-list designations that reach beyond classic asset freezes. Think of this as eight lists plus, breadth rather than a precise count. If a vendor screens two lists and calls the counterparty clean, the clean result only means clean against two lists. Ask which ones. Ask how often they refresh. Ask what happens to a query when a source is unreachable, because that returns you to the first test.

List breadth and honest reporting are the same problem viewed from two angles. Coverage decides how many checks the tool can run. Reporting decides whether the tool tells the truth about the ones it could not.

Why This Matters for EN590 and Refined-Product Trades

The documentary theater around a physical deal, the LOI, the ICPO, the DLC MT700, creates the impression of diligence. A stack of instruments is not a screen. The counterparty behind an ICPO can be a shell inserted into a mandate chain precisely because the buyer's screening will name-match the shell, find nothing, and move on.

EN590 gasoil and diesel cargoes attract this structuring because the tickets are large and the intermediary layers are normal in legitimate trade, which gives a dark fleet operator or a sanctioned beneficial owner cover to hide inside a routine-looking layer cake. In that environment, a screen that reports incomplete ownership resolution as clean is not a minor UX flaw. It is the exact failure mode the structure was designed to trigger. The MLRO who signs the trade is the one on the hook when the ownership chain is later unwound.

What Compliance Teams Should Do

• Run the pending test in every vendor demo. Ask the vendor to show, in live output, what the system returns when it cannot complete a check. Accept pending, unverifiable, or incomplete. Do not accept a clean pass on an unfinished check.
• Read the language of a pass. "No adverse match found" and "cleared against available data" are not the same as verified clean. Make the vendor define, in writing, what a green result asserts and what it does not.
• Confirm list coverage explicitly. Get the named lists, the refresh cadence, and the behavior when a source is unreachable. Expect the major consolidated lists plus sectoral and export-control coverage.
• Test the 50 Percent Rule path. Ask how the tool resolves aggregated beneficial ownership and what it reports when the ownership chain terminates before a natural person. That is where the layer cake hides.
• Make pending-not-clean a contract term. If the vendor's honest reporting behavior is real, it will survive being written into the agreement. If it will not, you have your answer.
• Document the false-negative posture in your vendor file. Under FATF Recommendation 10, you own the diligence. Recording how your screening tool handles incomplete checks is part of showing you understood the residual risk.

The buyer who leaves a demo able to describe a vendor's false-negative posture has done the work that matters. Demand pending, not clean. Confirm the lists. Everything else is decoration.

To see how a pending-not-clean design works in practice, book a walkthrough or subscribe to the OilFlow Intelligence briefing for evergreen typology breakdowns like this one.