Score distribution across the team
Where your analysts sit against each other and against the wider participant pool, by band rather than by league table.
For firms
See what your team detects, and what it does not
Every analyst works the same case. You receive the distribution of scores, the detection rate for each planted finding, the methods that were missed, and the unnecessary escalations.
- 8 minutes per case
- No integration
- Anonymous or named cohorts
- Method-level results
- Methods in the register
- 81
- Cite a regulator, court or agency
- 70
- Median age of cited evidence
- 9months
- Cases, across 11 sectors
- 23
The rubric is published, the register is open, and every method names the document behind it. You can audit the instrument before you buy it and before your third-party risk function asks you to.
The gap in existing AML MI
| Measure | What it tells you | What it does not |
|---|---|---|
| Alerts cleared per analyst | Throughput. | Whether the alerts were cleared for the right reasons. |
| SARs filed | Reporting volume. | How many were raised on customers whose explanation was already on file. |
| Training completion | That the module was watched. | That anything changed in how decisions get made. |
| QA sampling | Whether the process was followed and documented. | Whether the judgment underneath it was sound, since a well-documented over-escalation passes QA. |
Each of these measures activity. None measures whether the person doing the work can tell a genuine typology from a customer who looks irregular for a documented reason. The assessment measures that.
What you receive
Over-escalation rate
The proportion of your team's escalations raised on customers whose explanation was documented in the account notes, with a separate cut for those who escalated without opening the file at all.
Detection rates by finding
Which of the planted findings your team caught and which it missed, ranked by difficulty, so remediation has a target.
Sector benchmark
The same figures for the anonymised participant pool, so your distribution is read against something other than itself.
See one before you buy it
A worked example for a hypothetical cohort: the distribution, the detection rate for every method, what was missed, the over-escalation cut and the remediation order. It is produced by the same code as a real report. Only the scores are invented, and the document says so at the top.
Also available for Banking, Payments and EMIs, Trade finance, Capital markets and every other sector.
When firms use it
- 01
Training needs analysis
Per-finding detection rates show which typologies your team consistently misses, so the next session is aimed at evidence from this year.
- 02
Evidence of competence
A record that your firm assessed judgment and acted on the result. Supervisors ask for that when they want to see how competence is monitored.
- 03
New joiner baselining
A consistent eight-minute measure at week one, repeated at month six against a rotated scenario, which shows movement rather than tenure.
- 04
Calibration after a QA finding
When sampling flags an inconsistency between analysts, this shows whether the team is diverging on judgment or only on documentation.
- 05
Hiring
A candidate works a queue rather than describing how they would. The same scenario, scored the same way, for every candidate.
You choose what you see
You need to know who has completed and where to aim remediation. If analysts believe a score is going onto their appraisal, some will answer the way that protects them rather than the way they work. Attribution is therefore a setting, chosen per cohort, and whichever you pick is stated to the analyst before they begin.
The default is Confidential. You see who has completed and can chase the rest, and you see the team’s results in full. Individual marks are not attributed to a name.
Named
- Each analyst's score and band
- Who has completed and who has not
- Everything in the aggregate view
- Pick this when you intend to target remediation at individuals
Confidential (default)
- Who has completed and who has not
- Everything in the aggregate view
- No individual scores or answers
- Pick this when you want a true measurement and a chase list
A third mode, Anonymous, records no link between a run and a person, including for us. In every mode, no analyst’s written answers are shown to the firm, and distributions are suppressed below five runs. With three analysts, a band breakdown and one conversation identify who scored what.
£750. One cohort. One answer.
Put up to 25 people through the sector assessment and see which methods the team detects, which it misses and where it escalates unnecessarily.
- Full sector assessment
- Up to 25 participants
- Method-level detection rates
- Over-escalation rate
- Confidential cohort option
- No integration
- No customer data
- £750 credited against an annual licence within 90 days
Run a team pilot£30 per analyst at a full cohort
Need internal approval? Download the one-page business case
Measure the control, not the employee.
With a confidential cohort, management sees completion and team-level gaps without seeing individual scores.
Benchmark, remediate, re-test
- BenchmarkOne cohort, one sitting. The AML Judgment Gap Report names the methods the team missed and ranks them by how much exposure they carry.
- RemediateUsing your existing training and controls. We do not sell the remediation, which is what makes the measurement worth trusting.
- Re-testThe same methods on rotated cases, so the second sitting measures judgement rather than memory of the first.
- Measure improvementThe second report is read against the first, method by method, including anything that went backwards.
No integration · No customer data · Confidential cohort available
Security, data and procurement
- No customer data leaves your firm
- No integration and nothing to install
- Anonymous cohorts available
- No automated employment decisions
Read the data and privacy methodology for de-identification, retention windows, sub-processors and transfers.
AML Benchmark is a trading name of Net Werth Ltd, a company registered in England and Wales with company number 12718042. We are the UK data controller for the data described in the privacy policy, and where a provider sits outside the UK the transfer is covered by an International Data Transfer Agreement or an adequacy decision. For a data processing agreement, a completed due diligence questionnaire or your own supplier form, ask us.
Sit it yourself before sending it on
Eight minutes, the same scenario your team would work, with the full debrief at the end. Eight minutes is enough to judge the instrument. There is no demo to book first.
Start The Alert QueueFor a data processing agreement or a supplier questionnaire, ask us.