Employee Monitoring and Works Councils: Who Must Approve What?
Most employers who fall foul of works council rights on monitoring never intended to monitor anyone. They rolled out a ticketing system, a CRM, a security tool, or a productivity suite — and discovered, sometimes via an injunction, that in works council countries the legal question is not whether you intend to monitor employees. It is whether the system could. By that standard, almost every piece of modern workplace software is a monitoring system, and the approval landscape around it is one of the most underestimated topics in European employee relations.
Why "we don't use it that way" doesn't work
The instinctive defence — the logs exist but nobody looks at them — fails almost everywhere it is tested, because the major co-determination regimes attach rights to capability, not to use. A system that records who did what, when, and how fast can evaluate behaviour and performance whether or not anyone runs the report. Regulators reinforce the point from the data protection side: under the GDPR, and its Article 88 opening for member-state rules on employee data, the processing itself needs a lawful, transparent basis regardless of managerial intent. The result is a two-key lock on workplace technology in much of Europe: data protection compliance and collective approval, and unlocking one does not open the other.
Germany: co-determination with teeth
The strictest regime is §87(1)(6) of the German Works Constitution Act (BetrVG): the works council co-determines the introduction and use of technical devices designed to monitor employee behaviour or performance — and the Federal Labour Court has long held that objective capability is enough. Software that logs user activity, timestamps actions, or produces per-user statistics is in scope; German case law has pulled in everything from telephone systems to standard office and collaboration suites. Co-determination means the employer cannot lawfully deploy without agreement: the practical vehicle is a works agreement (Betriebsvereinbarung), and if the parties cannot agree, a conciliation board decides. A works council that discovers an unagreed system can seek an injunction to stop its use, and evidence gathered through it may be unusable in dismissal litigation. For group-wide tools, the central or group works council is often the right counterpart — one of many reasons to map your German bodies before the rollout plan is fixed. See the Germany country landscape.
The Netherlands: a consent right, not a conversation
Dutch law is nearly as demanding: under Article 27 of the Works Councils Act (WOR), regulations on employee monitoring facilities and on the processing of employee personal data require the works council's consent. Deploy without it and the works council can invoke the nullity of the decision; the employer's route around a refusal runs through substitute court approval, not around the council. Because the trigger is again capability — "facilities aimed at or suitable for observation" is the operative concept — HR systems, access controls, fleet telematics, and analytics dashboards all belong on the checklist. See the Netherlands country landscape.
France: consultation plus the evidence rule
In France, the CSE must be informed and consulted before the introduction of technology that permits monitoring of employee activity, and employees must be individually informed. The French twist is the evidence sanction: monitoring implemented without the required transparency has repeatedly rendered its fruits unusable against employees, even where recent case law has softened the absolutism at the margins. Combined with CNIL guidance on proportionality — permanent, comprehensive surveillance is presumptively excessive — the French rule is: consult first, disclose always, and design the monitoring narrowly. See the France country landscape.
The AI layer makes this bigger, not different
Every trend in workplace technology pushes deeper into this territory. AI systems that evaluate performance or allocate work are classified as high-risk under the EU AI Act, workers and their representatives must be informed before such systems go live, and emotion-inference at work is banned outright — we analyse that regime in detail in our AI Act and works councils guide. The point for present purposes is simpler: AI-driven tools inherit the entire monitoring approval landscape described above, plus the Act's own obligations. If your organisation's AI governance process does not have a works-council lane, it is missing a load-bearing step.
What a workable works agreement covers
Experienced negotiators converge on a familiar set of clauses: a defined purpose for the system and a ban on repurposing its data; who may access which data, for what, with what logging of the access itself; retention and deletion rules; an explicit position on performance evaluation — often a prohibition on using the data for individual assessment or dismissal; transparency towards employees; a change-management clause obliging the employer to return when the system materially changes; and a review rhythm. Employers who arrive with a credible draft covering these points shorten negotiations dramatically, because they have answered the questions the council was going to ask anyway.
Running this at multinational scale
The hard version of this problem is not one tool in one country — it is a global IT portfolio meeting a dozen national regimes at once. The organisations that manage it well maintain three artefacts: an inventory of systems with monitoring capability, mapped to countries and bodies; a register of the works agreements and consents that authorise them, with the change clauses that reopen them; and a consultation record that shows, for each deployment, who was engaged, when, and with what outcome. That is ordinary works council management applied to technology — and it is the infrastructure Graylark's works council management platform provides: bodies, agreements, consultations, and the audit trail in one system.
See how Graylark handles works council management