How to review a workplace investigation before relying on its findings

An investigation report may be complete in form and still leave an employer unsure whether it is safe to act on. A respondent may challenge impartiality. A board may find that the reasoning does not match the evidence. Human resources may discover that important material was never put to the person affected. Sometimes the report is unfinished, but a proposed disciplinary outcome is already being discussed. In each case the immediate question is not who would have been the ideal original investigator. It is whether the record now available is sufficiently fair, complete and reasoned to support the decision under consideration.

That is a distinct task from selecting an internal, external or hybrid investigator, which is addressed in our guide to when an employer should appoint an external workplace investigator. A review does not retrospectively change who conducted the investigation. It examines the work actually done, identifies the effect of any defects and recommends what further process, if any, is needed before reliance.

No single Australian statute prescribes a universal investigation or review procedure for every employer and every allegation. Applicable requirements may come from the Fair Work Act, Commonwealth, State or Territory discrimination and safety legislation, an enterprise agreement, contract, policy, public-sector framework or another sector-specific instrument. The appropriate review criteria therefore begin with the framework governing the matter, not with a generic checklist treated as law.

A review should respond to a reasoned concern, not merely to disappointment with an outcome. A bare assertion that an investigator was biased does not invalidate a report; equally, dismissing every challenge as dissatisfaction can conceal a real defect. The process owner should identify what is said to be wrong, where it appears in the record and whether it could have affected a finding or the proposed consequence.

• a challenge to impartiality based on prior involvement, reporting relationships, communications or treatment of participants
• allegations that were vague, changed during the process or lacked the particulars needed for a meaningful response
• adverse evidence, credibility concerns or a material version of events that was not put to the respondent
• an apparently material witness, document, recording or digital record that was not pursued, with no reason recorded
• interview notes that are fragmentary, disputed, unsigned or inconsistent with a recording, and no reliable account of what was said
• credibility preferences stated as impressions rather than explained by evidence, consistency, plausibility or reliable corroboration
• findings that exceed the appointment or terms of reference, or answer a different allegation from the one notified
• inconsistent use of the balance of probabilities, including language suggesting suspicion, possibility or a separate elevated standard
• communications or draft conclusions suggesting apparent pre-judgment before responses were received
• delay that may have degraded evidence, increased risk or made interim arrangements disproportionate
• a proposed serious disciplinary or termination outcome resting substantially on a report whose reasoning or process is contested

A desktop or quality-assurance review examines the existing record. The reviewer tests scope, process, evidence and reasoning but does not ordinarily gather new evidence or make new primary findings. It can say that the record supports reliance, that clarification is needed, or that identified gaps prevent a conclusion. It cannot reliably determine what an unasked witness would have said or reconstruct an interview that was never recorded adequately.

A supplementary investigation is targeted fact-finding intended to address defined gaps while preserving work that remains sound. It may involve obtaining a missing document, interviewing a newly identified witness, clarifying an ambiguous answer or putting adverse material that was overlooked. Once new evidence is gathered, the person doing the work must apply the same disciplines as an investigator: scope, fairness, reliable records and a clear explanation of how the new material affects each finding.

A fresh investigation starts again under a new appointment where defects are fundamental or confidence cannot reasonably be restored. That may be necessary where the original investigator had an incurable conflict, allegations and evidence cannot be separated from entrenched pre-judgment, the record is too poor to reconstruct, or the process investigated the wrong issues. Relabelling a reinvestigation as a ‘review’ does not change its substance. The engagement, communications with participants and resulting report should describe the work accurately.

The choice is not a hierarchy in which more work is always safer. Repeating sound interviews can burden participants, create inconsistencies through memory decay and prolong psychosocial risk. Conversely, a desktop endorsement cannot cure missing evidence by assumption. Select the least extensive intervention capable of producing a reliable basis for the decision that remains open.

An external reviewer is engaged and paid by the employer; that does not make the person structurally independent in every sense. The practical requirement is impartiality and freedom from relevant conflicts. Before appointment, test whether the reviewer advised on the complaint, designed or supervised the original process, coached a participant, reports to someone implicated in the matter, has expressed a concluded view, or may later be asked to defend their own earlier advice. Prior contact is not automatically disqualifying, but involvement in a question now under review may be.

The reviewer needs authority to identify defects without pressure to validate the report. The brief should name the appointing authority and recipient, the decision for which assurance is needed, the findings or process under review, the applicable instruments, the material to be provided, whether participants may be contacted, whether supplementary findings are authorised, the expected output and the boundary between quality assurance and legal advice.

Scope should also identify exclusions. A reviewer asked only whether procedural steps were documented should not imply that the evidence establishes the allegations. A reviewer asked to test findings should not decide sanction unless expressly appointed to advise on that separate issue. If the scope changes, record the variation and consider whether participants need to be told. Independence is not a label in the engagement letter; it is demonstrated through conflict checking, even-handed method and transparent reasoning.

The review should begin from a controlled file index rather than a selected bundle assembled to support the preferred outcome. Ordinarily the reviewer needs the complaint and intake records; the triage assessment and pathway decision; relevant policies, contracts, enterprise agreements and statutory or sector frameworks; the appointment and terms of reference; allegation letters, particulars and responses; investigation plans; interview invitations; recordings, notes, transcripts and participant confirmations; documentary and digital evidence; the working chronology; interim-measure decisions and reviews; correspondence with participants; drafts where they are relevant to scope, changes or pre-judgment; the final or partly completed report; and the material proposed for the employer's decision-maker.

A version history matters. It can show when allegations changed, whether a response caused genuine reconsideration, and why a draft finding moved. Drafts should not be demanded or circulated reflexively: their relevance, confidentiality and any claim of privilege require deliberate consideration. But where a reviewer is asked whether the reasoning was open-minded or whether material changes were explained, an unexplained gap between drafts and the final report may be material.

Record what is missing and what that prevents. Without reliable interview records, the reviewer may assess the report's internal logic but not confirm that it fairly represents oral evidence. Without the allegations sent to the respondent, the reviewer cannot determine whether notice was adequate. Without the governing policy or instrument, procedural compliance cannot be tested. An incomplete file does not always make every finding unsafe, but it narrows the conclusions available and should never be disguised by confident language.

Identify the employer, worker and location; the employment system; the nature of the complaint; and the decision contemplated. Then map the provisions that actually govern the process. An enterprise agreement may prescribe investigation or review steps. A contract may incorporate a policy. A public-sector code may impose procedural requirements. Safety, discrimination, whistleblower, privacy, surveillance or professional-regulation rules may affect particular evidence and disclosures. A procedure suitable for one workplace may be wrong for another.

For a dismissal within the national unfair-dismissal jurisdiction, s 387 of the Fair Work Act 2009 (Cth) identifies matters the Fair Work Commission must consider, including a valid reason related to capacity or conduct, notification of that reason, an opportunity to respond and any unreasonable refusal of a requested support person in discussions relating to dismissal. Those criteria are important where that jurisdiction applies; they are not a universal investigation code and should not be presented as one.

The reviewer should distinguish process compliance from factual reliability. A report may follow every step named in a policy yet reason poorly from the evidence. Another may contain sound factual analysis but depart from a mandatory instrument procedure. Both issues matter, but they call for different remediation and should be reported separately.

Each finding should map to an allegation or issue within the appointment. Compare the original complaint, triage record, terms of reference, allegation correspondence and final findings. Ask whether the allegation identified the conduct, approximate time and place, people involved and policy or obligation said to be engaged with enough clarity for a meaningful response. Precision is contextual; particulars need not reproduce every piece of evidence, but the person must understand the substance of the case to answer.

Scope can legitimately change when evidence reveals another incident or a different characterisation. The problem is silent expansion. New allegations should be authorised, formulated and put fairly before findings are made. A report should not convert a concern about one conversation into a broad finding about a pattern of behaviour without notice and evidence. Nor should it make adverse character findings that are unnecessary to answer the terms of reference.

Early intake and pathway selection are covered in our workplace complaint triage guide. At review stage, the narrower task is to determine whether the investigation remained aligned with the authorised questions and whether any departure affected the reliability or permissible use of its findings.

Fairness is practical, not ceremonial. The reviewer should ask whether adverse allegations and material were communicated in understandable terms; whether the respondent had a genuine and reasonable opportunity to answer before findings were concluded; whether relevant responses were investigated rather than merely recorded; and whether the investigator approached those responses with an open mind. The required process depends on the governing framework, the issue being decided and the consequences that may follow.

Not every discrepancy or adverse fact must be the subject of a separate letter. The question is whether the person knew the substance of the case that could materially support an adverse finding and could respond to it. Conversely, an allegation letter does not deliver fairness if the actual finding rests on a materially different event, policy breach or credibility proposition that the person could not reasonably anticipate.

A late opportunity to answer can cure some notice defects when it is genuine, provides the relevant material, allows reasonable time and occurs before a concluded decision. The response must then be considered with an open mind, and further inquiry may be required. It is not a cure if the outcome has already been irrevocably settled or the invitation is merely an opportunity to comment on a decision already made. Our broader guide to managing workplace complaints after triage addresses the end-to-end process; the review should focus on the particular defect and its effect.

Check whether potentially relevant records were identified early, preserved through authorised channels and collected consistently. Email, messaging, access logs, rosters, CCTV, document versions and device data may have short retention periods or legal restrictions. A report should explain significant unavailable evidence, efforts made to obtain it and how the absence affects confidence. Failure to collect every conceivable document is not necessarily a defect; failure to pursue an obviously material source without explanation may be.

For interviews, review the plan, sequence and method; the questions asked; whether the interviewer used leading or compound questions on contested matters; and how the account was recorded and confirmed. A signed statement is not the only reliable record, and a recording is not mandatory in every investigation. What matters is whether the record is sufficiently accurate for the use made of it and whether corrections or disagreements were preserved rather than overwritten.

Where notes are summaries, the report should not present them as verbatim quotations. Where a participant disputed a transcript or declined to confirm notes, the reviewer should inspect the disagreement and any recording rather than treating non-confirmation as proof that the account is unreliable. Trauma-informed practice may improve participation, but it does not predetermine credibility or displace fair testing of evidence.

Credibility is not a contest in confidence, seniority, presentation style or apparent emotion. The reviewer should identify why one account was preferred: contemporaneous records, internal consistency, consistency with objectively established events, plausible detail, admissions against interest, opportunity to observe, or reliable corroboration. Demeanour may be affected by culture, disability, trauma, language, neurodiversity or the stress of the process and should be treated cautiously.

Inconsistency requires analysis rather than a label. Some differences concern peripheral detail and may be expected with memory. Others concern the central event and require explanation. The report should test each account against independent material, identify evidence that cuts against the preferred view, and explain why it does not change the conclusion. Corroboration should be real: one witness repeating what another person told them is not independent confirmation that the event occurred.

A reviewer should not substitute a preferred conclusion merely because they might have weighed a close contest differently. The task is to decide whether the finding was reasonably open on a fair and adequate record and supported by a transparent chain of reasoning. Intervention is warranted where a material credibility preference is unexplained, rests on an irrelevant stereotype or overlooks evidence capable of changing the result.

Workplace investigators commonly determine whether an allegation is established on the balance of probabilities: whether it is more probable than not on the evidence accepted. The civil standard does not become a criminal standard because an allegation is serious. Nor does Briginshaw v Briginshaw create a third or elevated standard of proof.

The High Court's decision in Briginshaw concerns the strength and quality of evidence needed to reach reasonable satisfaction on the balance of probabilities in light of the seriousness of what is alleged and the consequences of the finding. Serious allegations should not be found on inexact proof, indefinite testimony or indirect inference where the evidence does not reasonably support satisfaction. That is an evidentiary discipline within the civil standard, not a percentage formula or presumption in favour of either party.

Review the words actually used. ‘Possibly occurred’, ‘could have happened’, ‘there is reason for concern’ and ‘I prefer the allegation’ do not necessarily express a finding on the balance of probabilities. Equally, reciting ‘Briginshaw’ adds nothing if the report does not analyse the evidence. The Fair Work Commission's guidance on valid reason relating to conduct explains that, where misconduct is relied on in an unfair dismissal matter, the Commission determines for itself whether the conduct occurred on the balance of probabilities. An employer's investigation report is evidence; it does not bind the Commission.

A reliable report makes its reasoning auditable. For each allegation it identifies the issue, governing definition or policy provision, evidence accepted and rejected, disputed facts, factual findings and ultimate conclusion. The conclusion should answer the allegation actually put. A long evidence summary followed by ‘substantiated’ is not reasoning; neither is a credibility conclusion repeated as the reason for itself.

Test materiality. If a factual error is corrected, does the conclusion change? If an omitted witness could only confirm an undisputed peripheral fact, the omission may not matter. If the finding depends on timing and the chronology is wrong, it may be unsafe. Separate defects that affect presentation from those that affect a finding, and findings that can be severed from those contaminated by the same reasoning or conflict.

Recommendations require particular care. An investigator may recommend systemic controls, policy changes or further assessment if the appointment authorises it. Recommendations should not trespass into the employer's disciplinary or termination decision by assuming sanction, undisclosed history or matters outside scope. The fact-finder establishes findings; the employer decides what action follows under the applicable legal, contractual and policy framework.

Reviewing consistency does not mean different matters must produce identical outcomes. It means checking whether comparable evidence was tested by comparable methods, whether one party was given procedural opportunities denied to another without reason, and whether departures from policy or the investigation plan were explained. Apparent bias may arise from prior involvement, prejudicial communications, selective evidence gathering or a closed mind; an adverse finding by itself is not evidence of bias.

Confidentiality should be expressed in qualified terms. Participants may be asked to limit discussion to protect privacy, evidence and safety, but an employer should not promise absolute secrecy or imply that lawful disclosures are prohibited. Information should be shared on a need-to-know basis, with access, storage, retention and disposal controls appropriate to its sensitivity. The Commonwealth Privacy Act employee-records exemption is limited and does not apply in the same way to every employer or every use. The Office of the Australian Information Commissioner's guidance explains its scope; State, Territory, health-record, surveillance and public-sector rules may also apply.

Review whether delay, communications, isolation, workload changes or interim measures created or aggravated psychosocial risk. The Safe Work Australia model Code on psychosocial hazards is authoritative national guidance, but model laws and codes have legal effect only as adopted in a jurisdiction. The employer should check the applicable WHS or OHS law and regulator guidance rather than presenting the model framework as uniform national law.

Engaging a lawyer, external investigator or reviewer does not automatically make the review, report or underlying evidence legally privileged. Marking documents ‘privileged and confidential’ is not decisive. Privilege is fact-specific and generally concerns confidential communications or documents made for the dominant purpose of obtaining or giving legal advice, or for use in existing or reasonably anticipated litigation. Operational fact-finding, quality assurance and disciplinary decision-making may have different purposes.

If privilege matters, obtain legal advice before settling the purpose, instructions, roles, reporting lines and information flows. Consider separately the status of source documents that existed before the review, communications seeking advice, the reviewer's working papers and any report. Also consider waiver: deploying a report's conclusions while selectively withholding the basis may create legal and fairness issues. AWS does not promise that an investigation review will be privileged and does not provide legal representation.

Review outcomes are practical descriptions, not legally prescribed categories. They should identify the scope examined, material available, defects found, effect on each finding and next step. A useful framework is:

• reliable — the reviewed record is sufficiently complete, fair and reasoned for the stated purpose, subject to the employer's separate decision process
• reliable subject to limited clarification — the conclusion remains open but a defined ambiguity, correction or explanation should be completed before reliance
• remediable by supplementary steps — identified gaps can reasonably be addressed through bounded evidence gathering or a genuine further response process
• unsafe to rely upon — one or more findings lack a sufficiently reliable basis for the proposed use and should not support action in their present form
• fresh investigation required — defects are so fundamental, pervasive or irreparable that confidence cannot reasonably be restored through supplementary work

Match remediation to causation. A transcription error may be corrected against the recording with the correction documented. An allegation lacking one particular may be clarified and answered before a conclusion. A missing witness may be interviewed and any new adverse evidence put to the respondent. An unexplained credibility preference may sometimes be repaired by reconsidering the complete record and providing actual reasons, but not by inventing reasons the original decision-maker never held.

Some defects cannot be reversed. Destroyed or overwritten evidence may leave a gap no interview can fill. An investigator's entrenched pre-judgment may contaminate choices about whom to interview and what to preserve. An incurable conflict can undermine confidence in the process as a whole. Long delay may make memories too uncertain to support a serious finding. In those cases the employer may need a fresh investigation, may need to narrow the issues capable of determination, or may have to accept that a finding is unavailable.

Cure is assessed before reliance and before a concluded employment decision. A later explanation drafted for proceedings does not repair the opportunity that should have been provided or recreate lost evidence. Record the defect, the chosen remedy, why it is adequate, who performed it and how it changed—or did not change—the analysis.

Where proposed action rests materially on disputed findings, pause reliance where appropriate until the review establishes whether the report is safe to use. A pause does not require all management action to stop. Preserve evidence immediately; maintain proportionate and reviewable interim safety or operational controls; identify any regulator, insurer or reporting timetable; and communicate realistic review stages without promising a date the work cannot meet. Our separate guide explains how to manage investigation duration and delay without inventing a fixed legal deadline.

Continue welfare and psychosocial-risk management for the complainant, respondent, witnesses and managers carrying the process. Avoid punitive or isolating interim measures, and monitor for adverse treatment after anyone makes or supports a complaint. Under the Fair Work Act and discrimination and safety regimes, retaliation, victimisation or adverse action can present issues separate from whether the original allegation is ultimately substantiated. The Australian Human Rights Commission's positive-duty guidance is relevant where sex discrimination, sexual harassment, sex-based harassment, hostile workplace environments on the ground of sex or related victimisation are engaged.

Keep roles intelligible. The reviewer assesses the investigation record; any supplementary fact-finder gathers and assesses authorised evidence; advisers explain options; and the employer's authorised decision-maker decides. Complete separation is not a universal legal requirement and may be impracticable in a small organisation, but conflating roles without recognising the resulting conflict or fairness risk weakens the process.

The reviewer's conclusion is not the disciplinary, termination, safety or organisational decision. The employer remains responsible for deciding what follows, through a person with appropriate authority and an open mind. That decision-maker should understand the review's scope and limitations, the status of each finding, any supplementary material, the employee's response to the proposed action, relevant history and mitigating circumstances, and every applicable legal, contractual, policy and industrial requirement.

A finding that conduct occurred does not prescribe sanction. Conversely, an unsafe finding cannot responsibly be converted into discipline through intuition or concern that something probably happened. Non-disciplinary controls may still be available where the evidence does not support an individual finding: work-design changes, policy clarification, training, supervision or risk controls, provided they are not used as disguised punishment and are otherwise lawful and appropriate.

Preserve a clean decision record. It should distinguish the investigator's findings, the reviewer's assurance conclusions, legal or workplace advice received, the employee's response and the decision-maker's own reasons. That distinction is especially important where the report contains recommendations, because accepting a recommendation without independent consideration may obscure who actually made the employment decision and why.

AWS can independently quality-review complete or partly completed workplace investigations, test whether the existing record supports reliance, identify defects that may be remediable and conduct supplementary fact-finding where that work is properly scoped. AWS also assists employers to design investigation frameworks, terms of reference, evidence and interview standards, review pathways and decision-role controls.

The scope is settled against the material available and the decision the employer needs to make. Where defects are fundamental, AWS can explain why a desktop endorsement is unavailable and what a fresh process would require. AWS is a workplace consultancy, not a law firm, does not provide legal representation and does not promise that an engagement will attract legal professional privilege.

This article is general information for Australian employers. It is not legal advice and does not account for the legislation, industrial instruments, contracts, policies, public-sector requirements or facts applying to a particular matter. Obtain legal advice where rights, privilege, disciplinary action, termination or regulatory exposure require it.