Workplace Bullying Complaints: How to Investigate Conflicting Accounts

One meeting, two sincerely different accounts

The manager recalls a straightforward conversation about a missed deadline. The employee remembers being humiliated in front of colleagues. Witnesses recall an angry tone but not the exact words. None of them has to be deliberately dishonest. Memory is selective, and the meaning assigned to a remark can differ sharply from the words spoken. A sound workplace bullying investigation therefore starts with questions about what can be established, rather than an assumption that one party must be lying.

In civil-style workplace investigations, practitioners often assess allegations on the balance of probabilities: is the event more likely to have happened than not? Australian guidance also invokes the Briginshaw principle, which calls for appropriately persuasive evidence when an allegation is particularly serious. Investigators look for contemporaneous notes, messages, witnesses and what each person said immediately afterwards. They reconstruct the relationship and circumstances without treating an interviewee's confidence, tears or body language as proof. They also guard against confirmation bias, the tendency to notice only evidence that supports an early theory. Detailed guidance is available, but high-quality experimental research on how workplace investigators decide between competing accounts remains limited. The Irish code on workplace bullying makes a useful starting point explicit: different people may perceive the same events differently.

Why recollections and perceptions diverge

Studies show how much a person's vantage point changes the picture. Depending on whether workers identified themselves, colleagues nominated them or both methods were used, one study found estimated bullying rates ranging from 3.9% to 39.6% in the same workplace group. Estimates of perpetrators ranged from 2.7% to 19.3% (Coyne et al., 2003). These are not interchangeable measures, and the spread does not tell us that any one witness was lying. It tells us that the method matters.

Our self-image also affects reporting. In another study, 5.4% of respondents acknowledged bullying others, while 8.3% identified themselves as targets (Matthiesen and Einarsen, 2007). Written communication adds a separate source of misunderstanding. In an experiment, people were 89% confident that they could read the tone of an email correctly, but their interpretations were accurate only 63% of the time (Kruger et al., 2005). An investigator should consequently ask for the precise message, the surrounding exchange and what each person understood, without collapsing the three into a single 'fact'.

How HR loses neutrality, and when mediation is the wrong tool

The HR team is not automatically neutral simply because it administers the policy. British research found that some HR practitioners approached complaints against managers with scepticism, partly because of pressure to protect the organisation (Harrington et al., 2012). Another study showed that even a well-designed anti-bullying policy can fail when line managers apply it poorly (Woodrow and Guest, 2014). Independence must be built into the process: whoever investigates should have no reporting relationship or other conflict of interest with either party.

Mediation is useful in some cases, but it is not a universal substitute for fact-finding. Under the Irish code, voluntary mediation can be particularly valuable early in a dispute about working relationships, with a suitably trained mediator and a clear boundary between mediation and a later investigation. UK guidance also allows mediation in some bullying disputes, though it is generally unsuitable as the sole response to serious disciplinary allegations. Research cautions that mediation by itself may leave the organisational causes of bullying untouched (Jenkins, 2011). A joint meeting should never be imposed on someone who fears intimidation or retaliation. The question is not 'Can we get them in one room?' but 'Is this a safe, voluntary and appropriate way to address this particular problem?'

What complaint procedures require in different countries

The legal details vary, so an international organisation needs local procedures rather than a single legal script. Poland: under the reform effective 5 November 2026, employers with at least ten employees have until 5 May 2027 to put the required anti-bullying rules and procedures in place, unless these matters are already addressed in applicable workplace rules or a collective agreement. The legislation does not prescribe one compulsory investigation committee; responsibility and reporting channels must nevertheless be clear (2026 amending act; Polish Labour Inspectorate). A trained, independent panel, including an employee representative where appropriate, can be a good organisational choice.

United Kingdom: the Acas Code of Practice sets expectations for informing people of allegations, allowing responses and relevant accompaniment, documenting decisions and offering an appeal. An employment tribunal can adjust an award by up to 25% for an unreasonable failure to follow the code where it applies. Ireland: formal bullying complaints should describe specific incidents; the civil service policy generally aims to complete investigations within three months, except in unusual circumstances. Germany: the General Equal Treatment Act, section 13 creates a complaint route for discrimination and harassment covered by that Act, with an obligation to examine and communicate the outcome. It should not be mistaken for a universal statute covering every form of bullying.

France: employers must respond appropriately to reports of possible moral harassment and protect staff; there is no single statutory template for every internal inquiry. Belgium: in a formal individual psychosocial intervention, a prevention adviser normally has three months to issue an opinion, with an extension possible in specified circumstances; subsequent employer deadlines depend on the procedure. Czechia: proposals to change how labour inspectors handle anonymous complaints were discussed in 2026, but should not be presented as a blanket rule already in force. Chile: the Ley Karin framework sets a short statutory investigation period, generally 30 days, and requires that both parties have the opportunity to be heard. The practical message is simple: check the rules in the country concerned before promising a particular deadline or appeal route.

Outcomes, reporting figures and individual needs

Every party should receive an appropriately detailed written outcome and information about whatever review or appeal route exists. An allegation not substantiated by the evidence is not the same as an allegation made maliciously. The Irish code draws that distinction. A careful outcome letter might say: 'We could not establish that the reported conduct met the relevant definition of bullying. This does not mean your account was fabricated or that the experience was unimportant.' An organisation should then check in after several weeks to see whether working relationships have stabilised and whether anyone has faced retaliation.

Complaint numbers are easy to misread. Some US estimates suggest that only 6–13% of people experiencing workplace bullying make a formal complaint; Polish survey findings indicate that 68% of affected people told nobody. Research on reporting sexual harassment points to the damage caused by an organisation's dismissive response or retaliation, rather than reporting itself (Bergman et al., 2002). More complaints after launching a credible procedure may therefore indicate greater trust, not worsening behaviour. As an analogy, research in English hospitals linked higher reporting of safety incidents with a stronger safety culture (Hutchinson et al., 2009). The number of cases should be considered alongside time to resolution, reports of retaliation and workplace climate, not in isolation.

Fairness also means making interviews accessible. Acas guidance on neurodiversity recommends considering practical adjustments rather than waiting for a formal diagnosis: questions in advance, a written response option, clear agendas, adequate breaks or an appropriate companion. A UK employment dispute involving an autistic employee illustrates the harm poorly adapted procedures can cause. Ask everyone what they need to give a clear account. Instead of guessing what a colleague meant, ask what they heard and understood. Directly seeking the other person's perspective can be more accurate than imagining it (Eyal et al., 2018).

A practical investigation checklist

Start by checking whether voluntary mediation is both safe and suitable; do not use it to sidestep serious allegations. For a formal process, write down the allegations, scope, standard of assessment, expected stages and a realistic deadline, such as 30 days with a reasoned extension where appropriate. Appoint an investigator with no conflict of interest and, where possible, a different decision-maker. Give both parties a fair opportunity to respond, keep accurate interview notes and allow them to check the record of their own statements. Offer a review by someone previously uninvolved where the applicable policy provides one. Document even a brief initial response to a report.

Ask every participant what would help them communicate effectively, not only those with a disclosed disability. Protect identities on a need-to-know basis and make it easy to report retaliation. Where a complaint concerns senior leaders, an independent external investigator may be more credible. At the end, distinguish 'not proved' from 'false', deliver a written decision and arrange a discreet follow-up. Finally, review procedure quality, fairness and safety rather than treating fewer complaints as proof of success.

Related reading: Why we misjudge relationships at work and Why witnesses of workplace mistreatment remain silent.

When accounts conflict, the quality of the questions, the range of evidence and the transparency of the conclusions all matter. A fair process lets each person speak, establishes what can be supported and explains the next steps. Mediation should be considered in light of the case and the willingness of those involved.

Empatyzer: preparing to speak to both sides of a complaint

An HR specialist is preparing to interview people who remember a meeting in completely different ways. She asks Em: “How can I ask questions without putting answers in their mouths?” Em could help her build a sequence: “What exactly did you hear?”, “Who was there?”, “What was agreed immediately afterwards?” It can then help organise notes into events, accounts and points still to check, and prepare a calm explanation of the next stage. She goes into the interviews with a plan that supports attentive listening and consistent treatment of both sides. An anonymised account is enough to prepare, and the conversation with Em remains private.