What a Good NRM Needs

Here is a test most European governments would fail. Take a person who has just escaped a trafficking situation. He is 18, speaks about 87 words of the local language, has no documents, and is terrified of the police. He was forced to deal drugs and on some days work on a building site.

Now ask whether your state's National Referral Mechanism will reliably identify him at first or even second opportunity, protect him, and help him rebuild his life, including through provision of long-term psychological assistance. In most of Europe, the answer is probably not. That is not a polemical claim. It is what two decades of GRETA evaluations, three rounds of Eurostat data collection, and numerous insights from research, workshops and reports show.

The legal framework, it must be said, is not the problem. The architecture is largely in place. What is missing is the ability, or in some cases the will, to build what the law already requires.

The binding law says more than people think

The Council of Europe Convention on Action against Trafficking in Human Beings (CETS No. 197, 2005) does not use the phrase "National Referral Mechanism." It does not need to. Its substantive obligations define everything a functioning NRM must deliver, article by article. Article 10 requires states to train competent personnel in identification and to ensure those personnel collaborate with support organisations. Article 10(2) goes further: where there are "reasonable grounds to believe" a person has been trafficked, that person must not be removed from the territory until identification is complete, and must immediately receive assistance. Article 12 explains what that assistance should include, amongst other things “Each Party shall adopt such legislative or other measures as may be necessary to assist victims in their physical, psychological and social recovery.” Article 12(6) contains one of the most important single sentences in anti-trafficking law: assistance shall not be made conditional on the victim's willingness to act as a witness. That sentence directly contradicts how some states organise their NRMs in practice.

The EU's Directive 2011/36/EU reproduced this architecture at EU level. Article 11(2) triggers support at the "reasonable grounds indication" threshold. Article 11(3) reaffirms that cooperation with prosecutions cannot be a condition of receiving help. Recital 25 lists the range of officials who must be trained: police, border guards, immigration officials, prosecutors, judges, labour inspectors, social workers, health and consular staff. The original Directive was deliberately vague about structure, requiring only "appropriate mechanisms" under Article 11(4). That vagueness was a mistake, and the legislatures have corrected it.

Directive (EU) 2024/1712, which entered into force on 14 July 2024 and must be transposed by 15 July 2026, amends the 2011 instrument and closes the gap. EU law now requires referral mechanisms; the Directive, paragraph 15 of the recital states an NRM is an essential measure to enhance cross-border cooperation. The recital then goes onto explain that “a referral mechanism should be a transparent, accessible and harmonised framework facilitating the early detection and identification of, assistance to, and support for the victims of trafficking, and facilitating their referral to the responsible national organisations and bodies.” As the ink dried on the pages, it became clear that Member States can no longer point to informal inter-agency arrangements and call it a day.

What the OSCE understood that governments still resist

The OSCE/ODIHR published the first edition of its NRM handbook in 2004, a year before the CoE Convention was even open for signature. The second edition, published in January 2022, remains the most operationally useful document in this field. Its definition of an NRM is worth quoting precisely: "a co-operative framework through which state actors fulfil their obligations to protect and promote the human rights of trafficked persons, co-ordinating their efforts in a strategic partnership with civil society."

Note what that definition contains and what it excludes. It does not describe a government database, a hotline, or a ministerial coordination committee. It describes a framework built on partnership, which means NGOs are structural participants, not contractors called in when needed. The OSCE's 2022 edition set out four pillars: identification and protection; individual support and access to services; social inclusion; and criminal justice and redress. Pillar three is the one most governments ignore. Intensive support in a crisis shelter is not the same as helping someone build a life. Long-term integration, housing, employment, and education are where nearly every European NRM falls short, and where nearly every victim eventually encounters the limits of state protection and is put at risk of re-trafficking.

The 2023 OSCE Guidance on Trauma-Informed NRMs added another dimension that European governments have been slow to absorb. Trauma-informed practice is not a therapeutic add-on. It is a precondition for effective identification. A victim who is re-traumatised during an identification interview is a victim the system will lose.

The persistent failures are not mysteries

I see a few key systemic gaps across all EU Member States: inadequate early identification because frontline staff are not trained and/or do not fully understand all the nuances of human trafficking; a structural bias towards identifying female victims of sexual exploitation that misses men, labour exploitation, and forced criminality; near-universal failure to enforce the non-punishment principle and thus failure to provide adequate protection; absent or unreliable monitoring; near-zero effective victim compensation, and systematic conflation of immigration enforcement with victim identification. Underpinning all this is a lack of a single codified NRM document with clear roles, clear accountability, clear timelines, and clear monitoring

What a good NRM actually requires

Strip away the procedural language and the answer is straightforward.

A functioning NRM starts before any victim even materialises. It requires mandatory, regular training for every professional who is likely to encounter a trafficked person: police officers, border guards, labour inspectors, healthcare workers, social workers, school teachers, and consular staff, to name just some examples. That training must cover three things at least, how to spot the signs, how to engage with a potential victim, and what to do next.

Once a potential victim is informally identified, the system must make it easy for them to enter it. That means a dedicated NRM website, available in multiple languages, carrying all relevant referral forms and referral submission space, contact numbers, and plain-language explanations of what happens next.

A functioning NRM then needs a low-threshold formal identification trigger, because if you wait for certainty before offering help, you will routinely help no one. Provisional support must begin the moment there are reasonable grounds to believe someone has been trafficked — this is not generosity, it is what Article 10(2) of the CoE Convention and Article 11(2) of the EU Directive both require. Formal identification, however, should be carried out by a dedicated body with actual expertise in human trafficking: a Multi-Agency Trafficking Identification Panel, composed of trained specialists operating under clear statutory timelines. If that Panel reaches conclusive grounds, it issues a formal victim status certificate. That certificate matters practically, not just symbolically. For example, a person with no identity documents who needs to open a bank account, access healthcare, or rent accommodation has no way to prove who they are or what has happened to them. The certificate is the bridge between legal recognition and daily life.

It then needs unconditional assistance, because conditioning support on cooperation with prosecutors is both legally non-compliant and operationally counterproductive: traumatised people do not give reliable evidence when they are simultaneously being threatened with removal. That assistance should be delivered by specialist NGOs, with guaranteed multi-year funding from the state, not project-by-project contracts that expire every twelve or eighteen months and require organisations to spend weeks preparing competitive bids instead of supporting victims. Multi-year funding matters because trafficking recovery is not a sprint. A survivor may need months of psychological support before they are ready to engage with legal proceedings, employment services, or housing options. Services need to exist in time for that. Multi-year funding also allows NGOs to build genuinely tailored provision: separate shelter for men, for women, for families, and for children.

An NRM needs a single national database, with standardised data entry across all agencies, so that one victim is counted once, the state has reliable figures, and policy decisions are made on evidence. The absence of a unified database is not a technical oversight. It is a political choice, and it is one that consistently allows governments to underestimate the scale of the problem they are failing to address.

Final thoughts

The legal framework is in place. The knowledge is available. The transposition deadline is July 2026. At some point, the question stops being "what does a good NRM need?" and becomes "why haven't we built one?"