The Transfer Test Is About to Stop Working

• European transfer law asks whether an economic entity retains its identity. In labour-intensive services, workforce continuity has often been central to that assessment; in automated services, that signal may become less reliable.
• AI-enabled delivery redistributes capability across people, models, platforms, licences and workflow logic.
• That creates a gap in two directions. Transfer protection may engage less often, while the capability actually moved sits outside employment law entirely.
• Leaders need a capability-continuity test beside the legal checklist. Nothing in the current or pending legal architecture supplies one.

Directive 2001/23/EC applies where an economic entity retains its identity through a transfer. Rights and obligations move automatically. The transfer itself cannot be grounds for dismissal, though dismissals for economic, technical or organisational reasons remain available. That carve-out is where most post-deal restructuring happens.

One correction, because it runs opposite to the post-Brexit assumption. The UK is not outside this family. TUPE 2006 transposes the Directive and adds a second, UK-only limb capturing service provision change: outsourcing, insourcing and re-tendering, independent of whether an economic entity transferred in the CJEU sense. Few mainland regimes offer an equivalent codified trigger.

Two consequences follow. Employee agency is not uniform: Germany allows a conditioned objection right, the Netherlands a qualified one, while France treats transfer as automatic and Belgium sits closer to automatic than to election. And variation rules differ materially by jurisdiction, so integration plans routinely assume redesign before the legal and social architecture permits it.

In Süzen, the CJEU held that losing a service contract to a competitor, without transferring significant assets or a major part of the workforce, is not a transfer. Factor weighting is sector-dependent: in labour-intensive activities, a stable group of workers can itself constitute the economic entity. Oy Liikenne is the mirror image. In bus transport, failure to take over the vehicles was decisive against finding a transfer even though most drivers moved.

Now apply the commercial shift. ISG reports BPO annual contract value fell 14% in 2025, the lowest since 2020, while underlying activity held steady, attributing this to automation compressing task-centric, seat-priced work. Everest Group estimates automation could replace 25 to 40% of FTEs in major BPO markets.

Put the two together.

Picture a claims-adjudication service transferring between providers. Five years ago: 200 people and a workflow tool. Today: 15 exception-handlers and an LLM-based adjudication model licensed separately by the client. The analysis increasingly turns on whether the model moves. If it does not, the argument that no transfer occurred gets stronger, and the 15 people holding every piece of undocumented operational judgement fall outside a regime designed to protect exactly them.

Two qualifications. No CJEU ruling has yet tested this, so it is a doctrinal prediction extrapolated from pre-AI authority, not decided law. And labour-intensive versus asset-reliant is a spectrum, so most near-term cases will be hybrid, the clean outcome reserved for near-fully agentic functions on skeleton teams.

Deal lawyers may point to TSAs and IP licences. Fair. But the problem is narrower and harder to dismiss: it is well-lawyered but poorly designed. Two teams, two timetables, two documents. Employment owns who transfers. Commercial owns what the service runs on. Neither owns whether operational capability survives, because that question belongs to no document. That is why the issue is not a missing clause; it is a missing architecture.

1. Contractual. Roles, terms, accrued rights. Moves by law. Everyone manages this layer.
2. Codified. Process documentation, runbooks, models, data, tooling, IP. Moves by contract and licence. Not by labour law.
3. Tacit. Judgement, sequencing, workarounds, what normal looks like on a bad day. Moves only if the individual transfers and stays.
4. Relational. Customer trust, regulator familiarity, credibility with the works council. Moves to nobody. It gets rebuilt or lost.

Cutting across all four: after transfer, who is authorised to say no? Decision rights are rarely transferred deliberately. They are inherited by accident, discovered by escalation, and settled 6 months late.

1. Which of the four layers move, and by which instrument? Name the document for each.
2. Which dependencies are operationally load-bearing but contractually invisible?
3. Is consultation only a compliance gate, or also knowledge capture? It is the one process that puts you in a room with the layer 3 holders.
4. Where does the variation constraint collide with the synergy case, in each jurisdiction?
5. If the service is automating, run the identity test yourself before someone else does. If the tooling does not move, do you still have a transfer? And if not, what protects the exception-handlers?

The acquired-rights regime protects people from being treated as disposable when businesses change hands. That principle is not weakening and should not.

But it was built for a world where transferring work meant transferring workers. That assumption is now carrying more load than it can bear, and the identity test is where the strain will show first.

What is striking is how little of this is measured. Transition-specific attrition, service degradation attributable to knowledge loss rather than process change, and the 12 to 24 month capability decay window all remain under-researched. Bloor is opening a research programme to close those gaps with primary data.

If you have run a multi-jurisdiction transfer since 2023, particularly one involving automated or AI-enabled delivery, I would like to compare notes.

SOURCES

Directive 2001/23/EC (EUR-Lex). TUPE 2006, reg. 3 (legislation.gov.uk). Süzen, C-13/95; Oy Liikenne, C-172/99 (CJEU).

WTW M&A readiness survey, reported June 2025. ISG Index Insider, February 2026, and 2026 State of BPO. Everest Group, November 2024.

This is analysis of operating-model design, not legal advice. Obligations differ materially by jurisdiction and should be assessed with qualified local counsel.

A fully source-backed version is available as a Bloor Research Note: Capability Continuity in Cross-Border Transfers.

Print Article for this blog Download here