Can HR see individual 360° feedback results? Who should see what – and why it decides whether the feedback is honest

The short answer: it depends on the access model

Any 360° software can be configured so that HR sees everything, nothing, or something in between. The honest answer is therefore: whatever your programme rules say – and if they say nothing, raters assume the worst. They answer the question for themselves before they click the first scale, and that answer decides how candid they are. If you do not fix and communicate the access model, it has still been fixed – just not by you.

DefinitionAccess model

The written rule stating which role – leader, supervisor, HR, consulting team, coach – may open which level of 360° data, and when. Three levels need separating: each rater's individual answers, the individual report with group averages and free text, and process or aggregate data such as completion rates or themes across a cohort.

The first level is uncontroversial: no role should be able to attribute an answer to a specific person, the consulting team included. How a system enforces that is covered in Is 360° feedback anonymous?. The arguments start with the report.

Who sees what: the access matrix

The matrix is the model we recommend for development programmes – not a law of nature but a decision, and every deviation belongs in the programme charter.

Two rows explain the whole conflict. Supervisors receive not the report but the development plan the leader derives from it. Bracken, Rose and Church (2016) count manager involvement and follow-up accountability among the conditions for effective 360° feedback – involvement, though, does not mean access to data that reads like an appraisal. HR receives process and aggregate data: enough to steer, nothing to judge with. What a report contains is described in What a 360° report should contain.

Three common models – and what each does to the feedback

In practice we meet three access models. They differ on one question: who owns the report?

The first model protects candour; the third protects it and adds follow-through. That is not a matter of comfort. In the meta-analysis by Smither, London and Reilly (2005) of 24 longitudinal studies, improvement after multisource feedback was generally small; it was more likely when recipients perceived a need to change, set goals and took action – with a coach or by discussing the feedback, as the authors put it. A report nobody discusses changes little; a report the wrong person reads changes the feedback itself. Which delivery model fits is compared in Consultant-led or self-service?.

Why development programmes should keep HR out of individual results

The argument has nothing to do with distrusting HR but with a structural dual role: the function that sets up development programmes also prepares promotion rounds, pay decisions and exits. Even if the individual HR person would never use a report for a decision, raters and leaders cannot know that – and people act on what is possible, not on what was promised.

Feedback research supplies the mechanism. Kluger and DeNisi (1996) showed across 607 effect sizes that feedback interventions improve performance on average (d = .41) but that more than one third made it worse – and that effectiveness declines as attention shifts from the task to the self. That is exactly what happens when a report turns into a personnel file: the leader reads it as a verdict on their worth, not as information about how they lead. Hensel and colleagues (2010) reach a related conclusion from their reliability study: 360° feedback works better as a development tool than as an administrative appraisal system.

What HR gives up is less than it seems:

• Steering: who has started, who has finished, which reminder is due.
• Evidence: whether and when every report was released and debriefed – for programme reporting and budget.
• Programme design: aggregate themes across the cohort – which dimensions are weak across the board, which patterns cluster – as the basis for training offers.
• Governance: assurance that anonymity rules, consultant review and access rights actually hold.

What happens when results feed promotion decisions

In short: from the second cycle on, the instrument measures not leadership but the game around the instrument. Raters whose scores affect a promotion round up for people they like and down for rivals. Leaders nominate sympathisers and avoid critical voices – how to catch that is described in How to select raters. Self-ratings turn tactical: anyone who knows the self–other gap will be judged describes themselves as they want to be seen. The result looks like a 360° and no longer is one.

There is also a legal dimension we only name here: under Article 22 GDPR, everyone has the right not to be subject to a decision based solely on automated processing that produces legal effects concerning them or similarly significantly affects them. A 360° score as the sole basis of an automated promotion decision would therefore fall under Article 22 – permissible only under one of its exceptions and with the safeguards it prescribes, such as the right to human intervention. The data-protection questions a programme must settle are collected in GDPR and 360° feedback.

How to write the access model into a programme charter

A programme charter is a one- to two-page document that precedes the first invitation and is identical for everyone involved – leaders, raters, supervisors, HR, the consulting team and, where one exists, the works council. Five points belong in it:

1. Purpose in one sentence. "This programme serves the development of the leader and is not used for promotion, pay or exit decisions." Without that sentence everything else is decoration.
2. Data levels, roles, timing. The access matrix, completed for your programme: when reports are released, when and above which cohort size HR sees the aggregate, whether supervisors get a column of their own – and whether they know beforehand.
3. Sharing by the leader. The leader may share their report – with their supervisor, their coach or nobody – and not sharing carries no disadvantage. The Article 29 Working Party held in 2017 that, given the imbalance of power, employees can give free consent at work only where refusal has no adverse consequence; the same yardstick applies to "voluntary" sharing.
4. The route for requests that come anyway. Who answers a partner's request for a manager's report, and with which sentence? A charter that does not cover this case breaks the first time it happens.
5. Retention, deletion, repeat cycles. Who deletes what, when and at whose instruction – a question for your data-protection function, not a vendor promise. In the second cycle the comparison with the first follows the same roles.

How LEADBeyond 360° handles it

As a concrete example of the third model: in LEADBeyond 360° the access matrix is a property of the platform, not a setting. HR and client admins draft cycles, add leaders, propose process settings, see which raters are assigned to each leader with group and status, send reminders, and see whether and when a report was released; every cycle is configured and activated by the consulting team. They see no individual scores, report content, PDFs, free text or response-quality indicators, and cannot change the anonymity threshold, scoring or release mode.

The model has a price we name openly: an HR-owned model in which HR reads and forwards every report cannot be configured in LEADBeyond 360° – and nobody should want it in a programme that promises development. Supervisor scores appear separately – like every group – only once the minimum number of responses is reached. Showing them with a single supervisor is an explicit per-assessment opt-in by the consulting team; it is off by default, and the supervisor concerned should know beforehand.