A 360 review becomes unreliable the moment one person’s opinion sets the frame for everyone else. When managers can see self-assessments before rating, or peers can see prior feedback, later responses tend to follow the strongest narrative rather than the evidence. Knowing how to run a blind 360 review gives HR and managers a practical control: each contributor assesses performance independently, against the same defined work.

The objective is not to remove judgement from performance management. It is to make judgement visible, comparable and accountable. A well-run blind process exposes meaningful gaps between self-perception, manager assessment and peer experience without allowing any one view to influence the others before submission.

What a blind 360 review should hide

Blind review is often used to describe two different controls. The first is blind submission: self-assessments, manager ratings and peer feedback remain locked until each assigned reviewer has submitted. This prevents premature influence and is particularly useful when comparing an employee’s view of their work with a manager’s assessment.

The second is rater anonymity: individual peer feedback is not attributed to a named person in the final view. This can encourage candid input, especially where peers depend on the employee or where team hierarchy makes open feedback difficult.

These controls should not be treated as interchangeable. A manager’s assessment generally should be attributable because the manager is responsible for performance decisions. Peer identities may be withheld, aggregated or shown depending on team size, culture and the purpose of the review. In small teams, promising anonymity when comments are easily identifiable can reduce trust rather than improve it.

How to run a blind 360 review

A blind 360 review works best as a defined operating process, not as an anonymous survey. Set the role criteria, reviewer groups, submission rules and calibration process before anyone receives an invitation. If those decisions are made midway through the cycle, employees may reasonably question whether the process has been adjusted to fit a preferred outcome.

Start with the actual role, not generic competencies

Generic criteria such as “leadership”, “communication” and “initiative” rarely produce comparable assessments on their own. They invite reviewers to apply their own definition of good performance. Instead, begin with the employee’s position description and translate it into the responsibilities, deliverables and behaviours that matter in that role.

For a customer success manager, that may include account retention, renewal planning, issue escalation, quality of account records and cross-functional coordination. For a people leader, it may include workforce planning, coaching, delivery oversight and decision quality. The criteria should reflect the work the employee is expected to perform, not a broad list of desirable traits.

AI-assisted role analysis can reduce the administrative burden here by extracting tasks and responsibilities from position descriptions. The HR team should still review the output. A position description may be outdated, overly broad or missing responsibilities that have become material during the review period.

Build a rating framework reviewers can apply consistently

Use a small, clear scale and attach observable definitions to each rating. Five scales are usually enough for most organisations. The key is not the number of points, but whether managers and peers understand what evidence separates one rating from another.

For example, a scale can distinguish between performance that requires regular support, performance that reliably meets the role requirement, and performance that improves outcomes beyond the expected standard. Include examples that relate to the role where possible. “Communicates effectively” is vague. “Provides timely, accurate handovers that allow other teams to act without rework” gives reviewers something they can assess.

Do not ask every reviewer to rate every criterion. Managers are best placed to assess overall accountability, delivery and role expectations. Peers can provide useful evidence on collaboration, handovers, technical contribution or customer impact. Employees should complete a self-assessment against the same core expectations, with space to identify context that may not be obvious to others.

Select reviewers for evidence, not popularity

A 360 review is not a vote. Choose reviewers who have sufficient exposure to the employee’s work during the review period and can comment on specific interactions or outputs. A balanced group might include direct reports, cross-functional partners, project stakeholders and peers, depending on the role.

Set a minimum number of peer responses before displaying aggregated feedback. This protects anonymity and limits the impact of a single unusually positive or negative response. The right threshold depends on team structure, but HR should avoid presenting peer data as a reliable pattern when only one or two people have contributed.

Also define exclusions. A reviewer who has worked with the employee only briefly, has an active interpersonal conflict, or cannot assess the selected criteria may add noise rather than useful evidence. Exclusion is not about avoiding difficult feedback. It is about ensuring difficult feedback comes from a credible source with relevant evidence.

Lock submissions until every required assessment is complete

This is the operational heart of a blind 360 review. Configure the system so employees cannot view manager or peer input before submitting their self-assessment, and managers cannot view the employee’s assessment before submitting theirs. Peer reviewers should not see other peer comments or ratings while completing their responses.

The lock should apply to both scores and written comments. If a manager can read an employee’s explanation before rating, even without seeing the self-score, the assessment is no longer independent. Likewise, an employee who sees a manager’s ratings first may shape their self-assessment around anticipated disagreement rather than their own evidence.

Set firm deadlines and send targeted reminders. Do not unlock the review simply because one party is late. If the process regularly stalls, address the accountability issue separately or establish a documented exception process. Unlocking early to keep the cycle moving weakens the control that makes blind comparison valuable.

CALIBRATE applies this side-by-side model by keeping self and manager assessments locked until both parties submit, then presenting the completed assessments for comparison. That sequence is simple, but it changes the quality of the conversation: both parties arrive with an independently formed view.

Ask for evidence alongside ratings

A rating without evidence is hard to calibrate and easy to challenge. Require concise comments that identify relevant examples, outcomes or patterns. The aim is not to turn the review into a case file. It is to give the manager and employee enough context to discuss why a rating was selected.

Useful prompts include: what work outcome supports this rating; what behaviour helped or limited performance; and what should the employee continue, change or develop next cycle. For peers, focus prompts on observed collaboration and impact rather than asking them to make employment decisions.

Reviewers need guidance on language as well. Feedback should describe work, decisions and behaviours, not personality labels. “Missed agreed handover dates on two project milestones” is actionable. “Not committed” is an interpretation that may conceal a workload, priority or process issue.

Calibrate the results before final decisions

Once submissions are unlocked, compare the assessments in a structured order. Start with the areas where self and manager ratings are materially different. These gaps are not automatically evidence that one party is wrong. They often reveal different access to information, unclear expectations, shifting priorities or a lack of feedback during the period.

Review peer input as supporting evidence, not as an automatic average. A low peer score may signal a genuine collaboration issue, but it may also reflect a project conflict, a process failure or a reviewer group that does not see the employee’s core work. Managers should test patterns against examples and role requirements before deciding on a final assessment.

For larger organisations, run a calibration discussion across managers after individual reviews are complete. Compare standards, not people. The question is whether similar performance has been assessed consistently across teams, levels and functions. HR should record any rating changes and the evidence behind them. A defensible process can explain not only the final outcome, but also how that outcome was reached.

Turn the review into a focused performance conversation

The review meeting should not begin with a manager announcing a score. Begin with the evidence, the areas of alignment and the most useful differences in perspective. A self-assessment that is higher than the manager’s may reveal achievement that has not been visible. A manager assessment that is higher may indicate the employee is underestimating their contribution or working to an outdated standard.

Finish with a small number of specific next actions tied to the role. If feedback identifies inconsistent stakeholder updates, define the expected update cadence, audience and quality standard. If the issue is capability, identify the work exposure, coaching or support needed. Vague development commitments create the same ambiguity the blind review was designed to remove.

A blind 360 review is most valuable when it becomes a repeatable discipline: independent assessment first, evidence-based comparison second, and clearer expectations for the work that follows.