Guide

360 feedback in a German organisation: where the works council comes in

If you’re planning structured feedback on named colleagues at a company with a works council, the first question is whether the council has to agree to it. The Betriebsverfassungsgesetz doesn’t mention 360 feedback, so the answer comes from provisions written for other things and from Federal Labour Court decisions on employee surveys, and this page sets out what those say, how a 360 run with me is built, and which questions I haven’t found an answer to.

General information, not legal advice. Statute texts and court decisions read on 3 November 2026.

The short version

Five things the law and the decisions say.

  1. Nothing in the Betriebsverfassungsgesetz is written for 360 feedback. A council’s rights attach to features of the set-up: the system the answers are collected in, the questions, whether taking part is voluntary, and who sees the results. Neither a round for one person’s development nor one run across a leadership population is exempt on its own terms.

  2. The provision most likely to apply is § 87 (1) no. 6, which gives the council a say over technical systems suited to monitoring employees’ behaviour or performance. In 2018 the Federal Labour Court read “suited to” objectively, and treated collecting the information as enough. What the employer intends doesn’t decide it.

  3. The same decision found no further rights over reworded questions in a voluntary survey where employees rated unnamed supervisors. A 360 asks about a named person, so that reasoning can’t simply be carried over. A 2017 decision on a voluntary, anonymous paper survey reached the same point about voluntariness.

  4. Data protection law applies separately. § 26 (2) BDSG makes the voluntariness of an employee’s consent depend on their dependence and the circumstances, and § 26 (4) allows collective agreements, such as a works agreement, to serve as the legal basis.

  5. I haven’t found a court decision on a 360 run by an external coach with raters the subject has chosen. Where that leaves things is set out below.

What a works council can claim

The provisions, one at a time.

Each card says what the provision covers, what it could mean for a feedback round, and what the Federal Labour Court did with it in the 2018 survey case. Where the council and the employer can’t agree, the conciliation board (Einigungsstelle) decides and its ruling replaces their agreement (§ 87 (2); § 94 (1)).

§ 87 (1) no. 6 BetrVG

Technical systems

Covers
Introducing and using technical devices that are designed to monitor employees’ behaviour or performance. The council decides jointly with the employer.
In a feedback round
A system that records how colleagues rate each other’s behaviour at work can fall inside it.
In the 2018 case
The survey system was a covered device, because it is objectively suited to collecting information about conduct or performance. It had already been agreed with the council, so reworded questions created no new right.
§ 94 (1) BetrVG

Personnel questionnaires

Covers
Forms that ask employees for personal information need the council’s consent.
In a feedback round
A 360 asks colleagues for views on a named person, and the subject’s self-rating is personal information too.
In the 2018 case
Not applied: taking part was voluntary, and the questions asked for subjective views on unnamed supervisors. The court didn’t address named subjects.
§ 94 (2) BetrVG

General assessment principles

Covers
General principles for assessing employees need the council’s consent.
In a feedback round
One set of criteria applied to every manager in an organisation looks more like a standing scheme. A round built around one person’s question looks less like one. A court hasn’t decided where the line falls for 360.
In the 2018 case
Not triggered: nothing set abstract criteria for a uniform, group-wide assessment.
§ 87 (1) nos. 1 and 7 BetrVG

Conduct and health

Covers
Rules on order and conduct in the workplace (no. 1), and health protection under the occupational safety rules (no. 7).
In a feedback round
These bite where a round sets rules for how colleagues behave, or works as a risk assessment.
In the 2018 case
Both rejected: the survey set no rules for how employees behave towards each other and wasn’t a risk assessment, even though some questions touched on health.
§ 80 (2) and (3) BetrVG

Information

Covers
The employer has to inform the council in good time and comprehensively, and give it the documents it needs for its tasks on request.
In a feedback round
A description of what the system collects and who sees it is something the council can ask for.
Experts
§ 80 (3) lets the council bring in an expert where it needs one for its tasks, by agreement with the employer.
In the 2018 case
An information document set out the data flow in detail, and the IT documentation fixed the smallest group for which a report is produced at seven.

Where a round spans several sites or companies, which council is competent depends on whether the matter can be dealt with locally. §§ 50 and 58 BetrVG deal with the general works council and the group works council.

The closest decision

An employee survey, decided in 2018.

Court
Federal Labour Court
Decided
11 December 2018
Reference
1 ABR 13/17
Concerned
A group-wide electronic employee survey

A logistics group ran an annual employee survey in an electronic system, a survey that had run every year since 2007 and that a third party conducted. The group works council had agreed the current version of the system in a group works agreement. For the 2015 round the employer reworded, added and dropped questions in the section where employees assess their direct supervisors, who aren’t named, and the council claimed a say over the new questions. In December 2018 the Federal Labour Court (case 1 ABR 13/17) found it had none.

Three points from the reasoning matter for a 360. First, the court looked at the system and not at the wording of the questions. The system was a technical device under § 87 (1) no. 6, and because the council had already agreed it, changed questions didn’t change what the system does. Second, the court dealt with anonymity in small groups: reports were produced only for groups of at least seven participants, yet in a unit with a single direct supervisor that supervisor could be identified, because statistical data on the participants was linked to the answers. The seven-person threshold was part of that system’s agreed design. The court didn’t approve it as a minimum for anonymity, and didn’t decide whether seven is enough. Third, voluntary participation and the subjective nature of the questions carried the § 94 analysis.

An earlier decision, from November 2017 (1 ABR 47/16), reached the same point about voluntariness. A hospital group had a third party supply and evaluate an anonymous paper questionnaire for its staff, and the court held it was not a personnel questionnaire under § 94 (1) because taking part was voluntary. It also held the survey was neither a health protection measure nor a risk assessment under § 87 (1) no. 7.

On § 87 (1) no. 6 itself, the court has read the provision widely. Entering data by hand and storing it can be enough, and the system only has to be objectively capable of capturing information about behaviour or performance. A law firm’s review of the case law lists the decisions on a social media page (1 ABR 7/15, December 2016) and on Office 365 (1 ABR 20/21, March 2022) as examples. In May 2023 the Federal Administrative Court decided two cases on public bodies’ social media pages under public-sector personnel representation law (5 P 16.21 and 5 P 2.22). It held that storing comments can count as monitoring, but asked whether there is actual surveillance pressure, which is a more demanding test than the objective reading above, and sent both cases back for further findings. They concern a different statute and not the BetrVG.

The 2018 decision says a survey system of that kind is within § 87 (1) no. 6. It doesn’t say what a council can demand of a system an external coach runs for one person, or of a questionnaire about a named colleague.

Where a 360 differs

Five differences from the survey, and what each leaves open.

AspectThe 2018 surveyA 360 run with meWhat it leaves open
Who is assessed Unnamed supervisors, reported as groupsOne named personWhether § 94 (1) applies to a questionnaire about a named person even when taking part is voluntary.
Who takes part All employees, invited to a group-wide surveyRaters the person assessed has chosenThe court’s voluntariness point may carry over. A court hasn’t said so.
Who gets results The employer, in aggregated reportsThe person assessed. I never pass a report to the organisation.The decision doesn’t say whether a council’s rights depend on who receives the results.
Who runs the system The employer, with a third-party providerI do. The organisation has no account.A third party ran the 2018 survey for the employer. Whether a coach engaged independently, with no employer commissioning or using the results, counts as introducing and using the system hasn’t been decided.
Purpose A group-wide survey of the workforceThe development of one personWhether a development round for one person is a “general assessment principle” at all. The court said the employer’s intent doesn’t decide the § 87 (1) no. 6 question.

None of the open questions in the right-hand column is answered by a decision I could find.

How the engagement is built

What I’ve designed, and what that settles.

The person assessed is my client, with their own account. They choose their own raters, who answer through a link and need no account. An area appears in the report only when it has at least three ratings, and that rule applies to everyone, me included. It is a reporting threshold, and it doesn’t guarantee that nobody could work out who said what. I never see a rater’s name or address, and I never see who was invited or who answered. Raters give ratings only, with no box for a comment. The person assessed sees who has answered but never what anyone said, and confirms before the round that the feedback isn’t used for employment decisions. The report is a PDF for the person assessed, handed over in a debrief, with my written observations alongside the results. There are no norms, percentiles or benchmarks. The organisation that commissions the round has no account, and I don’t pass it a report. If the person assessed shares theirs, that’s their decision.

This settles who receives results and what the report claims to measure. It doesn’t settle whether a council’s rights are engaged, because the Betriebsverfassungsgesetz binds the employer and I’m not the employer. That question is between the organisation and its council, and the design above is what I can put in front of both. I haven’t found a basis for saying the individual version sits outside the law either.

What a works agreement may need to settle

Seven questions that follow from the provisions.

The provisions above and the GDPR raise the same questions whenever a council and an employer agree on a feedback system. The list below comes from them and from no model agreement.

  1. Purpose The development of the person assessed, or something else. Under Art. 5 (1)(b) GDPR, data collected for one purpose can’t simply be used for another.
  2. Who takes part, and whether it is voluntary The § 94 analysis in 2018 turned on voluntariness, and § 26 (2) BDSG ties an employee’s consent to their dependence and the circumstances.
  3. Group size and identification The smallest number of ratings for which results appear, whether free-text comments appear, and what combinations of filters could reveal one person. In 2018 the number was seven.
  4. Who sees what The person assessed, a coach, a line manager, HR. This is where purpose limitation becomes concrete.
  5. Storage and deletion How long answers, reports and logs are kept (Art. 5 (1)(e) GDPR) and what happens to them when the agreement ends.
  6. Roles and agreements under data protection law Who is controller and who is processor, a written processing agreement where one party processes on another’s behalf (Art. 28 GDPR), and whether a data protection impact assessment is needed (Art. 35 GDPR).
  7. Term and review How long the agreement runs, how it can be ended, and how it is revisited when the system changes.
Data protection runs alongside

A second set of rules, with its own questions.

The works council’s rights and data protection law are separate. Ratings are personal data about the person rated, and they can be personal data about the rater too where an answer identifies them. § 26 (2) BDSG treats an employee’s consent with care because of their dependence, and § 26 (4) BDSG allows collective agreements to serve as the legal basis for processing employee data, provided they meet the safeguards in Art. 88 (2) GDPR. Whether a data protection impact assessment is needed is for the controller to assess under Art. 35 GDPR, before the processing starts. Where one party processes data on another’s behalf, Art. 28 GDPR requires a written contract, and whether that describes a coach and the organisation that commissions a round depends on who decides the purposes and means of the processing. A controller decides them, a processor acts on a controller’s instructions, and two parties who decide them together can be joint controllers (Art. 26 GDPR). The roles follow the actual arrangement and not who receives the report.

Which processors are involved is set out in the privacy policy, and this page makes no claim beyond it.

What stays open

Four questions I haven’t been able to answer.

  1. A system an external coach runs A third party ran the 2018 survey on the employer’s behalf. Whether a coach engaged independently, with raters chosen by the person assessed, counts as used by the employer under § 87 (1) no. 6 hasn’t been decided.
  2. A questionnaire about a named person Whether § 94 (1) applies when taking part is voluntary. The 2018 reasoning concerned unnamed supervisors.
  3. A development round for one person Whether it is a general assessment principle under § 94 (2) at all.
  4. The smallest group What number a council or a court would treat as enough. Seven was the figure in 2018, and the court didn’t rule on it.

The first three depend on how a court reads provisions written for other purposes, and no decision I found settles them.

Sources

Where each statement comes from.

Arrange a meetingHow a 360 works with me