Should Managers See Employee Survey Comments?

Managers usually should not receive raw comments from an anonymous employee survey. They need the meaning of the feedback: supported themes, how broadly each issue appears, what remains uncertain, and which organisational process deserves attention.

Verbatim access feels more authentic. It can also reveal an employee through a unique event, role, shift, project, or writing style. The result is often a worse report: leaders focus on identifying or debating one author instead of asking whether a pattern is real and what should change.

Why do managers want to read every comment?

The request is understandable. Scores show where sentiment moved, while written feedback can explain why. A manager may believe that reading the original words will provide nuance, urgency, or an unfiltered connection with the team.

Raw comments can certainly contain useful evidence. The question is not whether that evidence matters. It is who needs access to the original text to turn it into a sound organisational finding.

Giving every manager a comment export creates three common problems:

1. Identification becomes the hidden task. A manager recognises a project, meeting, phrase, or personal circumstance and starts guessing who wrote it.
2. A vivid sentence outweighs the pattern. One forceful comment can dominate ten quieter responses, even when it is unusual or factually incomplete.
3. The discussion moves from systems to people. Leaders debate the author's attitude, accuracy, or motive instead of examining the workload rule, escalation route, management practice, or resource constraint being described.

This is not an argument for ignoring free text. It is an argument for processing it before it reaches the people whose day-to-day knowledge makes recognition most likely.

How can a comment identify an employee without a name?

Names are only direct identifiers. Workplace comments also contain indirect identifiers: facts that become identifying when combined with what colleagues and managers already know.

Detail in a comment	Why it may identify someone	Safer reporting treatment
A recent customer, patient, or project	Only one employee handled it	Generalise the situation and report the underlying process
A shift, location, or unusual schedule	The team knows who worked at that time	Remove timing and location unless the pattern is broad
A rare role, qualification, or demographic	One person fits the combination	Roll up the group or suppress the cut
A manager conversation or complaint	The participants already know who was present	Summarise the issue without reproducing the event
Distinctive vocabulary or writing style	Familiar phrasing can be recognisable	Report a neutral theme, not a quotation
A precise date or sequence of events	Internal records can connect the comment to a person	Replace precision with a broader period or process stage
A response timestamp beside completion data	Operational records may narrow the possible author	Keep response content separate from delivery records

The Office of the Australian Information Commissioner explains that effective de-identification requires more than removing names. Organisations must consider unusual characteristics, what other information an audience holds, who will receive the data, and how likely recognition is in that environment. Its guidance specifically identifies providing analysis instead of raw data as a useful access control.

NIST Special Publication 800-188 makes the same broader point. De-identification requires a data-sharing model and risk assessment, not a single masking step. It also notes that unstructured text and dates need special attention because they can carry linkable detail.

An employee comment is unstructured text interpreted by an audience with unusually rich background knowledge. That is exactly why a sentence that looks harmless to an external reviewer may be obvious to a line manager.

What do managers actually need from written feedback?

A manager does not need every sentence to make a good decision. A useful finding should answer six questions.

Reporting question	What the report should provide
What is happening?	A neutral description of the supported theme
How much evidence supports it?	A count, range, or strength indicator that does not expose a small group
Where does it appear?	Only segments broad enough to remain protected
Why might it matter?	The likely operational, cultural, safety, or retention consequence
What remains uncertain?	Conflicting evidence, limited support, or facts the survey cannot establish
What should happen next?	A process, decision, or control to review, with an owner and follow-up date

For example, a raw report might present this sentence:

I raised the roster problem after last Tuesday's late shift, but my manager told me to stop complaining.

A findings-first report would not repeat it. If independent responses support the issue, it might report:

Employees describe inconsistent access to raise workload and scheduling concerns, with some believing that speaking up is discouraged. Review the escalation route, manager expectations, and how unresolved concerns move beyond the immediate supervisor.

The second version gives leadership something to test and improve. It does not pretend the survey proved every detail in one account. It also does not hand the manager a set of clues about the author.

Does summarising comments remove too much nuance?

Poor summarisation can flatten feedback. Good analysis preserves distinctions that matter while removing details that identify people.

The aim is not to turn every concern into “communication could improve.” A useful process should retain:

• the specific organisational mechanism involved;
• whether the evidence describes a repeated pattern or an isolated signal;
• important differences between groups, where those groups are large enough to report;
• the strength and consistency of the evidence;
• plausible alternative explanations; and
• what the survey can and cannot establish.

That requires judgement and quality control. It is still safer than distributing the source material widely.

The CIPD employee voice factsheet reports that a quarter of employees experience high levels of silence. It explains that people may withhold views when the risks of speaking feel greater than the benefits, especially when information challenges senior colleagues. A reporting process that makes recognition easy can reinforce exactly that calculation.

Research published in the American Economic Journal: Microeconomics also shows that survey security affects information transmission where people fear retaliation. Secure Survey Design in Organizations is a useful reminder that asking for truth and designing conditions for truthful reporting are different tasks.

What if a comment describes misconduct or a broken process?

An anonymous survey can uncover a serious signal. It is not designed to investigate or resolve a case for a particular unnamed person.

Suppose one response says a whistleblowing concern was never passed to the appropriate executive. The wrong next step is to search timestamps, compare writing styles, or question the team until the author becomes apparent.

The appropriate response is to examine the process:

1. Is the reporting route clear and available without going through the subject of a concern?
2. Is there a documented escalation owner and expected timeframe?
3. Can the organisation verify that reports reach the correct level without exposing reporters broadly?
4. Are managers trained on what must be escalated?
5. Have employees been told what changed after earlier process failures?

One anonymous response may be incomplete or mistaken. It can still justify checking whether a control works as intended. The organisation can later report that it reviewed the process, found a gap, and changed the escalation mechanism without confirming or denying an individual's account.

Where an investigation requires documents, questions, or individual support, employees need a separate confidential channel that is designed for case handling. That channel should exist before the survey. It should not be created by tracing someone who used an anonymous one.

When can verbatim comments be appropriate?

Verbatim access may be appropriate when the process is not anonymous and participants understand the model before responding.

Examples include:

• an attributed consultation where employees intentionally attach their names;
• a confidential case-management or speak-up channel with restricted investigators;
• a facilitated focus group with agreed notes and participant review; or
• explicit permission to publish a particular statement after the survey, obtained without pressure.

These are different mechanisms with different purposes. They should not be smuggled into an anonymous survey through a vague privacy statement.

For an anonymous survey, quotations are risky even when names and obvious details are removed. Vocabulary and sentence structure remain. A quote also invites leaders to interpret one person's tone rather than assess the weight of a theme.

The safer default is simple: employer reports should contain findings, not individual comments or lightly edited quotations.

How should comments become protected findings?

A findings-first workflow can preserve usefulness without granting broad access to source text.

1. Restrict the source material

Only the people or system responsible for analysis should access raw text. Line managers and general administrators should not receive a downloadable comment file “just in case.”

2. Separate identity and operational records

Invitation delivery, completion status, and reporting-group data may be needed to operate a survey. They should not become a route for joining a sentence back to a participant.

3. Remove more than names

Review events, dates, roles, locations, customers, colleagues, health details, and unusual combinations. Direct identifiers are only the first pass.

4. Code meaning consistently

Group comments around defined themes and subthemes. Keep positive, negative, and mixed evidence separate rather than forcing every sentence into a predetermined narrative.

5. Require support before reporting a finding

A finding should normally draw on multiple independent responses or align with protected quantitative evidence. A single comment may trigger a process check, but it should not be presented as established workforce sentiment.

6. Test the report for recognition risk

Ask what a motivated manager could infer using the report plus their knowledge of projects, schedules, absences, and conversations. Remove or broaden details that make one person stand out.

7. Preserve uncertainty

State when evidence is limited, mixed, or unable to establish cause. Privacy protection should not turn ambiguity into false certainty.

How should HR brief managers on survey findings?

Managers should receive a decision brief, not an evidence dump.

For each finding, ask:

1. Which process is implicated? Examples include workload allocation, rostering, promotion, manager escalation, psychological safety, or access to tools.
2. What can be checked without identifying anyone? Review rules, records, aggregate outcomes, training, handoffs, and control ownership.
3. What action is proportionate to the evidence? A weak signal may justify monitoring or an audit. A strong repeated pattern may justify immediate change.
4. What will employees hear back? Explain the finding, the action, the owner, and the next update date without exposing the source material.

This keeps attention on what the organisation controls.

How does Candora handle written feedback?

Candora does not show individual comments or raw responses to employers. When a survey closes, restricted processing turns written responses into supported, privacy-safe findings. Small groups are protected before employer reporting, and leaders receive the themes they can act on rather than the sentences they could trace.

There is no employer setting that silently changes this boundary. HR can manage invitations, completion, and reporting groups, but cannot open participant-linked answers or export a comment file.

The approach is designed around the purpose of an anonymous survey: to learn what patterns exist, where organisational processes may be failing, and what deserves further work. It is not designed to identify the person behind a difficult statement.

For a deeper look at the analysis process, read How Does AI Analyse Open-Text Employee Survey Feedback Safely?. The 30-day post-survey action plan explains how to turn protected findings into visible follow-through.

What is the practical answer?

Written feedback matters because it adds context that scores cannot provide. That does not mean every manager needs the original wording.

Give managers the theme, weight, operational context, uncertainty, and next decision. Keep individual sentences behind a restricted analysis boundary. When one comment raises a serious concern, inspect the relevant process without trying to discover who wrote it.

The right unit of action is the pattern, not the author.

See how Candora protects employee responses or explore a privacy-safe sample report.

Last source review: 24 September 2026. Privacy guidance and organisational practices change; confirm the linked sources before relying on them for legal or compliance decisions.