Reducing Sanctions Screening False Positives at a Wealth Manager

The Legal Backdrop in Brief

Since 15 June 2022, OFSI has been able to impose civil monetary penalties for breaches of financial sanctions on a strict liability basis. Section 54 of the Economic Crime (Transparency and Enforcement) Act 2022 removed the need for OFSI to show that a firm knew or had reasonable cause to suspect a breach, and OFSI’s enforcement guidance now treats knowledge only as a case factor. Wealth managers are relevant firms, so they must report to OFSI as soon as practicable when they know or reasonably suspect a client is a designated person, and they must apply OFSI’s ownership and control test. Since 28 January 2026 the UK Sanctions List has been the single source of UK designations, replacing the OFSI Consolidated List.

On the conduct side, the FCA expects systems and controls proportionate to your risk, with chapter 7 of its Financial Crime Guide (FCG 7) setting out good practice on sanctions. Its 2023 review and its May 2026 findings both looked directly at screening calibration, which is why tuning is now a supervisory topic rather than an IT one. For the wider regime, see our guide to sanctions screening and financial sanctions compliance.

Why a Wealth Manager Gets So Many False Positives

Name screening compares the names in your client records against names on the lists you screen. Because sanctioned persons use aliases, spelling variants and transliterations, systems use fuzzy matching and alert above a similarity threshold. Every design choice that catches more variants also catches more innocent people. The usual causes at a wealth manager are:

• Loose fuzzy thresholds: settings left at vendor defaults, or set low after a scare and never revisited. The FCA’s 2023 review found calibration problems in both directions: some systems were too sensitive and created excessive false positives, others missed simple name variations.
• Common and short names: frequently occurring forenames and surnames, and short names with few characters, produce high similarity scores against many list entries.
• Transliteration: names from Arabic, Cyrillic, Chinese and other scripts can be written in Latin characters in several ways, so systems allow wide variation.
• Poor client data: missing dates of birth, nationality or country of residence leave the analyst with only the name to work with, so nothing can be discounted quickly. Titles in the name field, joint names in one field and migrated legacy records all distort matching.
• Screening against too many lists: adding every list the vendor offers, without a reason linked to your client base, multiplies alerts with little benefit.
• PEP list overlap: where PEP and sanctions screening run in the same queue, PEP hits inflate the sanctions alert count and confuse two separate processes. PEPs are a due diligence question, not a sanctions one. See politically exposed persons explained.
• Related parties: screening trustees, beneficiaries, attorneys and beneficial owners is right, but it multiplies the names in scope, and their data is often thin.

Reducing False Positives Without Missing True Matches

The aim is fewer wasted alerts with no loss of detection, with every change tested and recorded.

1. Fix the data first

Data remediation usually has the most effect for the least regulatory risk. Report how many client and related party records lack a date of birth, nationality or residence, carry titles in the name field or hold two people in one record. Then remediate, highest-risk segments first. This sits naturally with periodic reviews and customer due diligence refresh.

Watch for placeholder values. The FCA’s 2026 findings noted dates of birth entered as placeholders rather than real dates, and that including honorifics and suffixes reduced match scores below alert thresholds. Bad data can cause false negatives as well as false positives, so cleaning it improves detection in both directions.

2. Use secondary identifiers properly

Once dates of birth, nationality and residence are reliable, they can be used to discount alerts quickly, or to auto-close clear mismatches under documented rules. For example, a rule might close an alert where both records hold a full date of birth and they differ materially. Rules like this need careful design: list entries often hold only a year of birth or several possible dates, and a rule that discounts on a placeholder date will hide a real match. Document each rule, test it against known true matches and have it signed off.

3. Tune fuzzy thresholds with documented testing

Threshold changes should follow evidence, not alert fatigue:

• Test file testing: run a set of known designated names, with realistic variants (transliterations, transposed name order, missing middle names, added titles, typing errors), through the system and record which ones alert at each candidate threshold.
• Above-the-line testing: review a sample of alerts just above the current threshold. If they are consistently false and the test file still alerts at a higher threshold, there may be room to tighten.
• Below-the-line testing: run the system at a lower threshold in a test environment and review a sample of the matches that would not currently alert. If true or near matches appear, the threshold is too high already.

Record the options considered, the results, the decision and who approved it. Repeat the testing periodically and after material list or system changes, which the FCA’s 2026 findings describe as good practice.

4. Good-guy lists, with periodic review

Once an analyst confirms a client is not a particular designated person, the system can suppress repeat alerts for that pairing. Keep it narrow: link a specific client record to a specific list entry, so a new designation or an amended entry still alerts. Review entries periodically and whenever either record changes. The FCA’s 2026 findings point to formal governance for approving and reviewing screening exclusions.

5. Scope the lists to your risk

Screening against the UK Sanctions List is the baseline for a UK wealth manager. Whether you also screen against US, EU or UN lists should follow your exposure: US person clients or staff, US dollar flows, EU clients or entities, and group requirements. Write the rationale down. Run PEP screening and adverse media as separate processes with their own thresholds and queues, so the sanctions queue holds sanctions alerts only.

6. Take a risk-based approach to settings

A segment of clients with links to higher-risk jurisdictions may justify more sensitive settings than a domestic retail book. Different settings are acceptable where the reasoning is documented, tested and supported by the business-wide risk assessment.

7. Triage, four-eyes and quality assurance

Level one analysts clear alerts with a written rationale using the secondary identifiers available. Possible matches go to a second reviewer, and confirmed or unresolved matches to the MLRO or sanctions lead. A quality assurance sample of closed alerts checks the rationales. Set internal service levels for each stage: the FCA’s 2023 review found significant alert backlogs worsened by weak governance and a lack of internal service levels.

8. Validate the model, including vendor settings

If you use a vendor or group platform, you still own the outcome: understand the matching logic, the settings, the lists in scope and how quickly updates load. Both FCA reviews found firms that could not explain their outsourced calibration, and the 2026 findings flag reliance on historic vendor settings. An independent validation should check the design, the data feed and the tuning evidence. See what you can and can’t outsource.

Governance: Documenting and Reporting Tuning

A tuning log should record each change, the reason, the testing before and after, the approver and the date. The MLRO or sanctions lead should sign off threshold, rule and list changes, and material changes should be reported to the board or risk committee.

The FCA’s 2023 review found that senior management did not always receive enough MI, and that some MI lacked basic metrics such as alert volumes and reports made to OFSI. Board MI on screening should normally cover:

The same material should feed the annual MLRO report.

Who Should Own Screening Tuning?

Accountability for financial crime controls usually sits with the MLRO (SMF17) or the senior manager responsible for financial crime. Day-to-day ownership of tuning works best with someone who combines sanctions knowledge with comfort handling data: a sanctions lead or a financial crime manager who can design a test file, read match scores, query client data and explain the results to the MLRO and the board. Operations and IT should implement changes, but they shouldn’t decide them.

At a smaller wealth manager the MLRO may own tuning directly, with periodic specialist support.

Signs you need a sanctions lead or data-minded financial crime manager

• Alert backlogs that recur after every major designation round.
• Screening settings that are still the vendor defaults, or that nobody can explain.
• No test file, no above or below-the-line testing and no tuning log.
• An internal audit or FCA finding on sanctions controls.

Our guide to hiring a sanctions lead covers the role in full, and our financial crime leadership recruitment service covers sanctions, MLRO and head of financial crime appointments. For a short tuning project, an interim appointment can work well.

Interview questions to test the skill

• How would you decide whether our current fuzzy threshold is right?
• Walk me through above-the-line and below-the-line testing. What would make you raise or lower a threshold?
• How would you build a test file for a client base like ours?
• What controls would you put around a good-guy list?
• Which lists should we screen against, and why?

Strong candidates talk about evidence, testing and documentation. Be cautious of anyone whose only answer is to raise the threshold. Our MLRO interview questions cover the wider financial crime judgement to test.

The Bottom Line

High false positive volumes are usually a symptom of poor data, untested settings and mixed processes. Fix the data, test thresholds in both directions, control exclusions, scope the lists and document every decision. If you need the person to lead that work, see our wealth management SMF recruitment page or tell us about your hire.

Frequently Asked Questions

Can we just raise the fuzzy matching threshold?

Only with evidence from test file and below-the-line testing showing you would not miss real matches, documented and signed off.

Is a good-guy list acceptable to the FCA?

Yes, if it is controlled. Suppressions should be narrow, approved, logged and reviewed periodically, and a new or amended list entry should still alert.

Should PEP screening sit in the same queue as sanctions screening?

It is better kept separate. PEP status is a due diligence matter, and mixing the two inflates sanctions alert volumes and blurs the escalation route.

Which list must a UK wealth manager screen against?

The UK Sanctions List, which has been the single source of UK designations since 28 January 2026. Other lists depend on your exposure, such as US persons, US dollar flows or EU clients.