Someone with ten years in know your customer (KYC) work wrote in an open forum that the job is data entry with extra steps. The most upvoted reply agreed and went further. The role is not there to fight crime, it said. It is there because finance treats the cost as a licence fee.
I have read some version of that argument for years. Parts of it are correct, and I will say which parts.
I sell anti-money laundering (AML) training. So I have an obvious commercial interest in the answer to this question being yes. I have tried to write this as if I did not.
One rule governs the whole article. Every number carries its source, scope, period, and publication date. Two of the most quoted figures in this field are routinely stripped of all four. Fixing that is most of what this page is for.
The case that AML compliance does not work, stated properly
Key point: the sceptical case is evidence-based, and its strongest parts cannot be answered by anyone honest.
The sceptical case against AML deserves to be put at its best, with no rebuttal attached. Rebuttal comes later, or this section is worthless.
Recovery is low. Almost no serious source claims otherwise. The Financial Action Task Force (FATF), the international standard setter for AML and countering the financing of terrorism (CFT), said so itself in 2022. I quote it below.
Most reports go nowhere visible. Europol reported in its 2017 report From Suspicion to Action that on average just over ten per cent of suspicious transaction reports (STRs), called a suspicious activity report or SAR in United States terminology, submitted to European Union financial intelligence units were put to use. The other reports were not necessarily worthless. But they produced nothing an analyst would recognise as an outcome.
The system is measured on inputs. Report volume, alert closure rates, training completion, file turnaround. All of these are counts of activity. None of them is a count of harm prevented. Supervisory inspections largely examine whether the machine ran, not whether it achieved anything.
Analysts almost never learn what happened. You file, and the file closes on your side. That silence is a design feature of the reporting regime, and I deal with it further down. It is also the single biggest reason people in this job stop believing it matters.
Does AML exist to protect the bank, not the public?
Partly yes, and the true part cannot be argued away: a bank's AML programme is built to protect its licence as much as the public. The claim that AML exists to protect the firm rather than the public is the strongest version of the critique, and it deserves a straight answer.
Part of it is simply true. A bank builds an AML programme because it must hold a licence, because supervisors inspect it, and because failures carry fines and personal consequences. Those are institutional motives. They are not the same as crime prevention, and pretending otherwise insults the reader.
What follows from that is narrower than the argument usually claims. A control built for licence protection can still generate intelligence. The institution's motive does not determine the usefulness of the output. But it does shape what the institution measures, and what it measures is where analysts feel the work go dead.
Money laundering statistics: where the most quoted figures actually come from
Key point: the two most repeated statistics in this field are usually quoted with the wrong scope, the wrong date, or no source at all.
This article exists to trace the most repeated AML effectiveness statistics back to their primary sources. I opened each source and read the sentence containing the figure, rather than the secondary summary that quotes it.
Three bodies are involved. UNODC is the United Nations Office on Drugs and Crime. The IMF is the International Monetary Fund. GDP means gross domestic product.
Table 1 puts the claim as it is usually stated in the first column and the source's own wording in the third. Read those two columns against each other before anything else. In every row, the published sentence is narrower than the claim built on it, and the final column names precisely what each figure does not support.
The same mistake produces all four distortions. A number gets separated from its scope and its date, and then it travels. A correct 2011 statement about global illicit financial flows becomes a 2026 statement about all criminal proceeds ever recovered, and almost nobody repeating it has opened the page it came from.
| Claim as usually stated | Primary source | What the source actually says | Scope and period | Published | What it does not support |
| "Less than one per cent of criminal proceeds is ever seized" | UNODC, Estimating illicit financial flows resulting from drug trafficking and other transnational organized crimes, page 5 | "Less than 1 per cent of global illicit financial flows are currently seized and frozen" | Global, illicit financial flows, data year 2009 | October 2011 | It is not a 2026 figure, and it is about flows seized and frozen, not about all criminal proceeds ever recovered |
| "Only one per cent is recovered" | Europol, From Suspicion to Action, page 30 | "just 2.2% of the estimated proceeds of crime were provisionally seized or frozen, and only 1.1% of the criminal profits were ultimately confiscated at EU level" | European Union only, 2010 to 2014 | 2017 | It is not global, not current, and the seizure figure and the confiscation figure are two different measures |
| "Two to five per cent of global GDP is laundered" | UNODC overview page, and the IMF range it rests on | UNODC's page publishes the range with no source and no year. UNODC's own 2011 study, page 7, calls the range the IMF's "consensus range", "made public in 1998" | Global, estimate of an unobservable quantity | 1998 range, restated since | It is not a measurement, and the page most people cite gives no year for it |
| "Effectiveness is failing worldwide" | FATF, Report on the State of Effectiveness and Compliance with the FATF Standards, page 11 | "In nearly all cases, effectiveness trails behind technical compliance." | 120 of 205 jurisdictions assessed | April 2022 | It measures assessed national systems, not the output of any individual firm or analyst |
The "less than 1 per cent of criminal proceeds" claim, traced to UNODC 2011
"Less than 1 per cent of global illicit financial flows are currently seized and frozen"
Read the scope. "Currently" meant 2011, and the study's monetary estimates are built on 2009 data. The report also says at page 5 that "the final monetary estimates are to be treated with caution. Further research and more systematic collection of data on this topic are clearly required."
Europol's figures are different numbers measuring something narrower. In From Suspicion to Action: Converting Financial Intelligence into Greater Operational Impact, published in 2017, section 9.2 on conversion rates, at page 30:
"Europol findings show that from 2010 to 2014, just 2.2% of the estimated proceeds of crime were provisionally seized or frozen, and only 1.1% of the criminal profits were ultimately confiscated at EU level."
European Union only. 2010 to 2014. Two separate measures, not one. Anyone quoting 1.1 per cent as a current global recovery rate is quoting a European figure from a period that ended over a decade ago.
Both figures point the same way. That is not the point. The point is that a compliance profession which insists on evidence in its own files should not repeat a number whose scope it has never checked.
The "two to five per cent of global GDP" claim, traced to the IMF 1998 range
UNODC's money laundering overview page states that the estimated amount of money laundered globally in one year is two to five per cent of global GDP, or 800 billion to 2 trillion United States dollars. Checked on 30 August 2026, that page gives no source and no year for the estimate.
The lineage is traceable. UNODC's own 2011 study says at page 7 that before its work, the most widely quoted figure for the extent of money laundering was the International Monetary Fund's "consensus range" of between two and five per cent of global GDP, "made public in 1998". The 2011 study's own central estimate was that the amount available for laundering through the financial system was "equivalent to some 2.7 per cent of global GDP", around 1.6 trillion United States dollars in 2009.
So the range is nearly thirty years old, it estimates a quantity nobody can observe, and the page most commonly cited for it does not date it. Use it if you must, but say all three of those things in the same sentence.
Technical compliance versus effectiveness: what FATF actually measures, and the eleven FATF Immediate Outcomes
Key point: effectiveness is not one question, and FATF says plainly that it is a different thing from technical compliance.
The distinction between technical compliance and effectiveness is the heart of this subject, and almost nobody explains it to practitioners.
Technical compliance asks whether a country's laws and rules match the FATF Recommendations. Effectiveness asks whether the system produces results. The FATF Methodology defines it directly:
"For the purposes of this Methodology, effectiveness is defined as 'The extent to which the defined outcomes are achieved'."
And it states the separation:
"Assessing effectiveness differs fundamentally from the assessment of technical compliance."
| Immediate Outcome | The question underneath it |
| 1. Risk, policy and coordination | Does the country know what it is exposed to? |
| 2. International cooperation | Does information and evidence cross borders? |
| 3. Supervision of financial institutions | Are supervisors doing real supervision? |
| 4. Preventive measures by firms | Do firms actually apply controls and report? |
| 5. Legal persons and arrangements | Can you find out who owns a company? |
| 6. Financial intelligence | Is intelligence used by investigators? |
| 7. Money laundering investigation and prosecution | Are cases brought? |
| 8. Confiscation | Is criminal property taken away? |
| 9. Terrorist financing investigation and prosecution | Are financiers prosecuted? |
| 10. Terrorist financing preventive measures | Are terrorist funds stopped from moving? |
| 11. Proliferation financing sanctions | Are proliferation funds stopped from moving? |
Look at what that list does to the question. "Does AML work" collapses into eleven separate questions, each with its own answer.
FATF's April 2022 Report on the State of Effectiveness and Compliance with the FATF Standards, covering 120 of 205 jurisdictions, gives the aggregate verdict at page 11:
"In nearly all cases, effectiveness trails behind technical compliance."
On confiscation specifically, at page 37:
"Consequently, only a small fraction of all proceeds of crime is currently being recovered and convictions for money laundering are often not in line with the major risks identified within each country."
That is the standard setter saying the system underperforms on the outcome the public cares about most. It is also, usefully, a system that measures itself publicly and publishes the results.
What FATF grey list monitoring actually changes, and how countries exit the grey list
FATF's grey list is the clearest published evidence that AML systems change state behaviour. FATF describes jurisdictions under increased monitoring as countries "actively working with the FATF to address strategic deficiencies in their regimes to counter money laundering, terrorist financing, and proliferation financing", which have "committed to resolve swiftly the identified strategic deficiencies within agreed timeframes".
Countries exit. On 23 February 2024 FATF stated in its Jurisdictions under Increased Monitoring statement that "The UAE is therefore no longer subject to the FATF's increased monitoring process", following work that included additional resources for its financial intelligence unit and increased money laundering investigations and prosecutions.
Whatever you think of asset recovery rates, this is an observable change in state behaviour, publicly assessed and published.
What happens to a suspicious activity report: the seven stages from filing to outcome, and where the chain breaks
Key point: there are seven stages between your report and a person being stopped, and the chain can break at every one of them.
What happens to a suspicious transaction report after it leaves your desk is the mechanical question underneath the philosophical one. The chain below uses the United Arab Emirates framework as the worked example, because that framework is written down clearly.
You form a suspicion. Under Federal Decree-Law No. 10 of 2025, Article 18(1), a firm that suspects or has reasonable grounds to suspect that a transaction or funds "represent Proceeds or are related to or intended to be used in the Crime, regardless of their value" must "notify the Unit without delay and directly, by providing a detailed report containing all available data and information".
You file it. Cabinet Resolution No. 134 of 2025, Article 18(1)(a), requires submission "through the Unit's electronic system or any other means approved thereby".
The Financial Intelligence Unit receives it. Article 11 of the Decree-Law establishes an independent Financial Intelligence Unit, or FIU, within the Central Bank, to which all STRs "shall be submitted thereto, exclusively".
It is analysed. Cabinet Resolution No. 134 of 2025, Article 46(3), splits this into operational analysis, to "identify specific targets such as persons, funds, or criminal networks", and strategic analysis, to "identify Crime trends and patterns".
It is disseminated, or it is not. Article 46(6) provides for dissemination to law enforcement "where sufficient grounds exist to suspect a link to the Crime". That condition is where most reports stop.
Law enforcement acts, or does not. Resourcing, priority and evidential standards all bite here. Intelligence is not evidence, and converting one into the other is expensive.
A case is brought, and property is confiscated. This is Immediate Outcome 8, and it is where the published assessments are weakest.
The chain breaks most often at stage five. Europol's ten per cent conversion figure measures that break point in the European Union between 2010 and 2014.
Stages one and two are the only ones you control. What you write at stage two is what a financial intelligence unit analyst reads at stage four, often months later and with no access to you, which is why our guide to writing a SAR narrative that survives triage is the most practical thing on this page.
Why you never hear what happened to your suspicious activity report
Two reasons, and only one of them is defensible. Tipping-off rules legally bar case information from flowing back to you. Structured feedback, where it exists, often isn't built through to the analyst who wrote the report.
The defensible one is confidentiality. Tipping off is prohibited. Cabinet Resolution No. 134 of 2025, Article 19(1), bars disclosing to a customer or any other person that a report has been made or that an investigation is under way. Case information cannot flow back freely.
The indefensible one is that structured feedback is often simply not built. Where it is built, it is a legal duty rather than a courtesy. Cabinet Resolution No. 134 of 2025, Article 46(4), requires the Unit to:
"Provide Financial Institutions, DNFBPs, and Virtual Asset Service Providers with feedback on reports received by the Unit to enhance the effectiveness of Crime combating measures and the detection and reporting of Suspicious Transactions"
Article 55(3) of the same Resolution requires law enforcement authorities to receive and follow up on the Unit's reports and to "provide feedback to the Unit regarding actions taken in connection therewith". The same duty appears in the primary statute at Article 9(3) of Federal Decree-Law No. 10 of 2025, which requires law enforcement authorities to "provide the Unit with feedback concerning the procedures taken in respect thereof". DNFBP means designated non-financial business or profession.
So the loop exists in law. Whether it reaches the analyst who wrote the report is a separate question, and in most firms it does not. If you want one thing to ask for in your next team meeting, ask what feedback your firm receives and where it goes.
For completeness, the protection that makes reporting possible is Article 37(1) of Federal Decree-Law No. 10 of 2025. No criminal, civil or administrative liability attaches to reporting entities or their employees for furnishing required information, "even if they were not fully aware of the nature or actual occurrence of the Crime, unless it is proven that the reporting was made mala fide with the intent to harm others".
What a well-written SAR narrative actually changes
Key point: quality doesn't change whether a crime happened. It changes whether your report survives triage and connects to anything.
Report quality is the narrowest claim on this page, and I won't stretch it.
A financial intelligence unit receives more reports than it can develop. Something has to decide the order. A report that states the suspicion plainly, names the parties, gives dates and amounts, and explains why the activity is inconsistent with what is known about the customer is cheaper to assess than one that does not. A report that requires an analyst to reconstruct your reasoning costs time that the queue does not have.
The second effect is linking. Your report may matter mostly because of what it connects to. A name, an account, a counterparty or a corporate vehicle that already appears in an open case turns a routine filing into a contribution. You will usually not know which of your reports did that.
Illustrative hypothetical. This describes no real person, customer, case, firm or report.
Two analysts file on the same trading company in the same week. Report A says the customer received third-party payments inconsistent with its stated business and attaches the statements. Report B says value and volume exceeded expected activity, cites the alert rule, and attaches nothing.
Report A can be triaged in minutes, and its parties can be matched against existing holdings. Report B has to be worked up before anyone can decide whether it is worth working up. In a queue, that difference decides order.
If you want the practical version, we cover how to write a report narrative that passes review separately. In the United States, the equivalent filing is called a suspicious activity report (SAR).
Predicate offences and the three stages of money laundering: placement, layering and integration
Money laundering is a secondary offence. It attaches to proceeds of something else: fraud, corruption, drug trafficking, tax crime, trafficking in human beings, sanctions evasion and others. FATF Immediate Outcome 7 assesses whether money laundering offences are investigated and prosecuted in line with a country's identified risks.
One clarification, because the two chains get confused. The money laundering stages taught in every induction course, placement, layering and integration, describe how criminal funds are moved. The seven-stage chain set out above describes what happens to your report. They are separate sequences, and effectiveness is measured on the second one, not the first. A firm can be excellent at spotting the stages and still sit at the wrong end of a reporting chain that stops at stage five.
I am not going to use any of those offences as an emotional device. The honest statement is mechanical. Financial reporting sometimes contributes evidence to cases about those offences, the contribution is usually partial, and the reporter usually does not see it.
System effectiveness and your effectiveness are different questions
Key point: a system can perform badly in aggregate while individual work inside it still determines individual outcomes. Both are true at once.
The argument for separating system effectiveness from your own effectiveness is short.
The aggregate evidence says national systems underperform on confiscation and prosecution. That is FATF's own published finding, not a critic's. It does not follow that no individual file matters, because aggregate underperformance is compatible with wide variation inside it.
It also does not follow that your work definitely matters. Anyone telling you it always does is selling something. What follows is narrower. Where a report is read, its quality changes what happens next. Where a control is designed and tuned by someone who understands the risk, it produces different alerts from one that is not. Where judgement is exercised, outcomes differ.
Whether your role contains any of that is a question about your role, not about AML compliance as a field. Some jobs are genuinely queue-clearing with the judgement removed. If yours is one of them, the critique is not wrong about your situation. It is wrong only when it is generalised from your situation to the whole field.
What is genuinely broken in AML compliance, and what would fix it
Key point: four things are broken, and none of them is the individual analyst.
Volume as the measure. Counting reports rewards filing, not filing well. Defensive reporting is a rational response to a system that penalises missed reports and never penalises useless ones.
Feedback absence. The loop is written into law in some jurisdictions and does not reach the desk in most firms. This is a solvable information problem, and it is misread daily as evidence that the work is pointless.
Input measurement. Supervisors can examine whether a control exists and operates. Measuring whether it reduced harm requires data most firms never see. So the examinable thing becomes the managed thing.
Judgement removed from the role. Where alert handling is broken into steps narrow enough to be scripted, the judgement leaves the job. That is a large part of what is actually moving offshore in AML, and it is also why the same complaint appears in teams that never offshored anything.
What would help is outcome measurement, structured feedback to reporters, and supervisory attention to whether reporting is useful rather than merely timely. None of that is inside an analyst's control, and I am not going to pretend a blog post fixes it.
What this means if you work as an AML analyst, and whether AML compliance jobs are worth staying in
Key point: decide whether your specific role contains judgement, then act on the answer.
Ask three concrete questions about your own AML role. They test whether the job contains judgement, challenge and feedback.
Does anyone challenge my decisions, and do I have to defend them? Does my firm receive feedback on its reports, and can I see it? Does someone read my written work and tell me why it came back?
If the answer to all three is no, the problem is your role or your firm, not the profession. Moving is a reasonable response. So is deciding this is not for you, and if that is your conclusion, I would rather point you to exit options from AML and what each one preserves than argue you out of it.
If you are staying, the useful things are unglamorous. Read your own quality assurance (QA) comments as data. Read published enforcement actions, because enforcement documents show what supervisors actually found rather than what training decks claim. Understand what your working day is supposed to contain, which we set out in what an AML analyst actually does all day. And keep an eye on which parts of the role are being automated, which we cover in whether AI will replace AML compliance jobs.
We sell AML training, so here is our interest in this
Key point: we have a commercial interest in this answer, and the specific claim we are and are not making is stated below.
Pro AML Training sells applied AML and CFT training. That is a direct commercial interest in you concluding that this skill is worth building.
So I will state what I am not claiming. I am not claiming that training improves national effectiveness outcomes. No one can source that claim, and it is exactly the self-serving statement this page should not make. I am claiming something much smaller: report quality affects triage and linking, and quality is a skill.
Judge the argument on the Europol page, the UNODC page, the FATF report and the Decree-Law articles quoted above. Not on who is making it.
If you want the mechanism this article calls its best evidence, our course on adapting to FATF grey list updates is free. If you want the judgement side, AML Gatekeepers in Action covers professional judgement in gatekeeper roles. Or Explore Courses across the catalogue.
How this article was researched and verified
Four primary documents were opened and read on 30 August 2026: the Europol report From Suspicion to Action, the UNODC 2011 research report Estimating illicit financial flows resulting from drug trafficking and other transnational organised crimes, UNODC's money laundering overview page, and the FATF Report on the State of Effectiveness and Compliance with the FATF Standards of April 2022, together with the FATF assessment Methodology. We read every quoted sentence in the source document, not from a summary.
The UAE provisions are quoted from the published text of Federal Decree-Law No. 10 of 2025 and Cabinet Resolution No. 134 of 2025. Article numbers are given because they were read, not inferred.
Frequently asked questions about AML effectiveness
Key point: fourteen direct answers on AML effectiveness, recovery rates, what happens to suspicious activity reports, and whether AML compliance is worth doing as a career.
