
Worker Protection Act 2023 · ACAS Code · Module 5
What happens in an investigation room shapes what happens in a tribunal room. The record is everything.
You are Claire Davies, Senior HR Investigator. Kelmar Group has retained you on a 15-day mandate to investigate the formal grievance filed by Tom Elliott against Diane Pritchard, Regional Director. You are external because Diane's seniority rules out internal investigation. The report you produce will sit on Kelmar's reasonable-steps file, will be read by the respondent, and — if this reaches tribunal — will be read by a judge.
Kelmar HQ, Manchester
A small interview room at Kelmar HQ. A wooden table, two chairs on one side and one on the other, a water jug, five folders stacked on the corner. Editorial framing. No dialogue.

The mandate is clean. The first material decision of the investigation is not a question of truth — it is a question of order.
Tom Elliott (complainant), Diane Pritchard (respondent), Stuart Henshaw (line manager witness), Rachel Adeyemi (HRBP witness). Eight working days available.
The ACAS Code does not prescribe an order. Procedural fairness, evidential coherence, and the respondent's right to be heard on specific allegations combine to give one strong order, two defensible ones, and one weak one. Each interview informs the next. The order you choose is itself a procedural record.
Tom on Tuesday, Stuart and Rachel on Thursday and Friday, Diane the following Wednesday. When Diane walks in she is asked to respond to specific allegations framed by the evidence you have gathered — not to a generic summary.
For Tom: his account goes on record before it can be coloured by other interviews. Procedurally clean.
For Diane: the specific allegations put to her are evidenced, dated, and precise. She can address each on its own facts — which is what natural justice requires.
For the organisation: the ACAS Code describes this sequencing almost verbatim as procedurally clean.
The sequencing is about evidential load-bearing. The complainant's account frames the allegations; witnesses corroborate or qualify; the respondent's interview is the moment at which specific, evidenced allegations are put to them for a specific, evidenced response. Reversing this order denies the respondent the thing that protects them most — precision.
Diane walks into Tuesday morning's interview and is asked to respond to Tom's written account before any witness has been interviewed. She answers at length, locking in a detailed counter-narrative. By the time Tom is interviewed two days later, Diane's version is the frame every subsequent question is implicitly testing.
For Tom: the investigation is now framed by the respondent's narrative. He will be interviewed against her account, not the other way round.
For Diane: her counter-narrative is on record before she has had to account for specific evidenced facts. The absence of documentary challenge at this interview protects her position.
For the organisation: the investigation's procedural weight is compromised. The Burchell standard of ‘reasonable investigation’ is weakened.
Respondent-first interviewing is not unheard of — but it is contextually weak in a harassment investigation where the complainant's evidence carries the initial load. Natural justice protects the respondent's right to be heard on specific allegations; interviewing her before those allegations are fully evidenced gives her less, not more, protection.
Stuart on day 2, Rachel on day 3, Tom on day 5, Diane on day 8. When Tom walks in on day 5 he is interviewed with witness accounts already on the record — which both sharpens follow-up questions and lets you test consistency across accounts.
For Tom: a defensible order. His interview is sharper because witness accounts are in first.
For Diane: the final interview is well-evidenced; specific allegations can be put precisely.
For the organisation: a less common but procedurally defensible order. The ACAS Code permits any order that meets the fairness tests. Scores slightly below complainant-first because witness recall of second-hand disclosure is typically weaker than first-hand account.
There are scenarios where witness-first makes sense: where the complainant is unavailable, where documentary evidence needs witness authentication first, or where witness accounts may be at risk of contamination if delayed. On this fact pattern, none of those apply — but the order is still procedurally defensible.
Everyone interviewed in the same week. You synthesise on Friday. The schedule works; the sequencing benefit is gone. Stuart, interviewed Tuesday, has not yet had the benefit of hearing Tom's account framed in your follow-ups. Diane's interview Friday is informed by Tom's Wednesday — but not by the witnesses, whose accounts are not yet transcribed.
For Tom: his interview occurs at roughly the normal point, but without the sharpening effect of waiting for witnesses.
For Diane: interviewed last in the week, which is sequencing-sound, but at the cost of evidence synthesis quality.
For the organisation: the parallel schedule saves calendar time at the cost of evidential coherence. A defensible choice where time pressure is severe; suboptimal where 15 days is the mandate.
The ACAS Code does not prescribe sequential interviews; it prescribes fair ones. Parallel scheduling is not a breach. It is, however, the scheduling pattern most likely to produce a report with follow-up gaps.

You are about to highlight the transcript. Six phrases are visible. Each needs a category tag: direct allegation, corroborating detail, contradiction with written statement, hearsay, opinion or character, or off-topic.
The investigator's craft is not deciding who is telling the truth. It is sorting each utterance into what it can do evidentially. Most weak reports treat every sentence as equally weighty — and collapse under cross-examination for it.
Categories: Direct allegation · Corroborating detail · Contradiction · Hearsay · Opinion / character · Off-topic. Over-tagging plain facts as ‘direct allegation’ carries a false-positive penalty.
Tom's interview · transcript pages 4–7

Five phrases from Stuart's interview need tagging. The weight distribution is different from Tom's — Stuart's direct observation is Monday conversation, not Friday night.
Stuart's account is a blend of direct observation (Monday conversation), second-hand information (Friday offsite), and opinion. The category sorting matters more here because the temptation to over-weight any of it is high.
Stuart's interview · transcript pages 2–4

The contradiction is live, documentary, dated, and on Kelmar's Teams server. How you handle the next ninety seconds will define the interview.
The investigator's task when documentary evidence contradicts a witness is to present the evidence, ask for a response, and record the response — nothing more.
For Tom: the contradiction is now in the investigation record, with Diane's explanation also on record.
For Diane: her right to be heard on specific evidence is preserved.
For the organisation: the handling meets the Burchell ‘reasonable investigation’ standard precisely.
The investigator's task when documentary evidence contradicts a witness is to present the evidence, ask for a response, and record the response — nothing more. Aggression removes the witness's ability to explain; silence denies the witness the chance to explain. Calm presentation preserves both the evidence's weight and the respondent's rights.
For Tom: the investigation is now delayed, legally contested, and any finding adverse to Diane will be challenged on the basis of investigator bias.
For Diane: her position is strengthened — the investigator's handling gives her grounds to challenge the process.
For the organisation: the procedural weight of the report just dropped materially. Uplift exposure rises if this reaches tribunal.
‘Hostile examiner’ is not a technical legal term — but the behaviour it describes is the specific thing a respondent's solicitor will cite as evidence the investigation was not impartial. Natural justice includes the rule against bias; investigator behaviour is part of that rule.
You move on. The interview completes at 12:20. Diane leaves believing her account has been accepted. Your report, filed eleven days later, cites the Teams message as a contradiction and bases a specific finding on it. Diane sees the report for the first time on appeal.
For Tom: a finding based on a contradiction Diane was never given the chance to address is vulnerable on appeal.
For Diane: her right to be heard on a piece of evidence that materially affects a finding has been breached.
For the organisation: appeals on process regularly succeed independent of the merits.
Natural justice requires that a person be given the opportunity to respond to evidence that is being used against them. Documenting a contradiction in the investigator's private notes, then citing it in a report, without putting it to the respondent, is one of the specific procedural failures that reliably overturns a finding on appeal.
For Diane: she is given the chance to clarify in her own words first. When the document is put to her, she has already framed her recollection.
For the investigation: tactical but procedurally fair. The witness is not cornered; she is invited to volunteer detail that either resolves the apparent contradiction or confirms it.
For the organisation: a defensible, careful handling. On balance, a solid procedural choice.
There is a defensible technique in which an investigator invites detailed clarification before presenting contradictory documentation. The line between this and ambush is whether the witness is given the chance to address the document in their own words once presented.
The role is reversed: the respondent is questioning the investigator. This happens in real investigations when the respondent is senior, legally advised, or both. Holding the line is the skill being tested — not winning the exchange.
Each round: pick the response. Score is based on procedural accuracy under pressure.
The report's structure is the final procedural decision of the mandate. It determines how the findings land with Kelmar, with Diane, with Tom, and — if this goes further — with a tribunal reading it three years from now.
The investigator's role ends at the finding. Per-allegation findings on the balance of probabilities is what the Burchell test anticipates.
The report goes to Lena Forsyth on day 14. Six allegations, six specific findings, civil-standard reasoning for each. The recommendations section proposes procedural next steps (disciplinary hearing, with options open) without naming a specific sanction. Lena reads it twice and signs it off for onward process.
For Tom: each allegation has an individual, reasoned finding.
For Diane: findings are at the right civil-law standard and specific enough for her to understand, appeal if she chooses, and respond to at a disciplinary hearing.
For the organisation: a report that meets the Burchell ‘reasonable investigation’ standard precisely, separates investigator role from decision-maker role per the ACAS Code.
Workplace investigations operate on the balance of probabilities — more likely than not. Single-verdict structures blur which allegations were supported and which were not. Per-allegation findings are not just clearer — they are what a tribunal will read first.
Your report delivers a single summary finding. Kelmar's disciplinary panel cannot tell which specific allegations were supported. The panel instructs a supplementary clarification; two days are added to the process. Diane, on receiving the report, files an initial appeal on the basis that the finding does not set out what she is found to have done.
For Tom: the report is directionally correct but unusable without re-work.
For Diane: the generic finding is harder to accept as fair — she cannot see specifically what she is found to have done.
For the organisation: the single-verdict structure imports a criminal-standard framing onto a civil-standard process. ERA 1996 s98 asks only whether the employer acted reasonably.
Workplace investigations are not trials. Findings are on allegations, not on persons, and are on the balance of probabilities, not beyond reasonable doubt. Single-verdict reports read as criminal-court drama transplanted into HR.
Your report includes per-allegation findings and a named recommended sanction: dismissal. The disciplinary panel, which is meant to make that decision, now opens its first meeting with an investigator's recommendation already on the record. The panel's independence is visibly compromised. Diane's eventual appeal leans heavily on this point.
For Diane: her right to a fair hearing before the decision-maker is weakened.
For the organisation: the investigator-role/decision-maker-role separation the ACAS Code treats as foundational has been collapsed.
For Tom: a successful appeal on procedure would mean the process has to be re-run — or the dismissal cannot stand — even if the underlying findings remain sound.
The ACAS Code separates the fact-finding function from the decision-making function for a reason: the decision-maker should approach the sanction question with an open mind. An investigator who names a sanction in the report has pre-empted the decision-maker.
Your report delivers per-allegation findings with full reasoning and no recommendations section at all. The disciplinary panel receives a clean findings document and builds its own procedural path forward. Kelmar's HR team asks on day 15 whether the report might include a ‘next procedural steps’ paragraph; you decline, citing the scope of the mandate.
For Tom: clear findings, usable directly at the disciplinary stage.
For Diane: a report that sets out what has been found and leaves the sanction question to the correct forum.
For the organisation: a defensible minimum position. Slightly less useful than c3a because the procedural next-steps guidance is absent, but not incorrect.
There is no requirement that an investigation report include a recommendations section. The minimum defensible report simply delivers findings. Both are procedurally sound.
Three decisions across fifteen working days. Two transcripts highlighted. One hot-seat test held. A report that either anchors Kelmar's reasonable-steps defence or weakens it — not on the basis of what was found, but on the basis of how the finding was arrived at.
1. Order is a procedural record in its own right. Complainant-first, respondent-last is the anchor sequence; other orders are defensible but carry specific evidential costs.
2. Evidence has weight classes. Direct allegations, corroborating details, hearsay, opinion, and off-topic material are not equivalent. The transcript highlighter is the craft layer — what separates a report that survives cross-examination from one that collapses under it.
3. Procedural fairness under challenge is the test. The hot seat is what a respondent's solicitor sounds like. Holding ground without becoming defensive, conceding on notice, referring questions upward where appropriate — this is the skill that protects both the investigator's independence and the respondent's right to be heard.
4. The investigator's role ends at the finding. Per-allegation findings on the balance of probabilities is what the Burchell test anticipates. Recommending sanctions collapses two distinct procedural roles; delivering verdicts imports the wrong standard.
Module 6 opens eighteen months later. The report you just produced is Exhibit J in the tribunal file. Jo Merrick and Lena Forsyth are reading it for the last time before the hearing.
Continue to Module 6 → Replay Module Course Home