How to run an investigationthat survives scrutiny.

When an investigation is criticised — at appeal, at tribunal, or in a grievance about the grievance — the criticism is rarely that the investigator reached the wrong conclusion. It is almost always procedural: the scope kept expanding, the notes were inadequate, it took seven months, the investigator was not independent, or the respondent was not told what they were actually accused of until the hearing.

These are all avoidable, and they are avoidable at the start rather than the end. Here are the eight points that, in our experience, determine whether an investigation holds up.

1. Write terms of reference before you begin

The single highest-value document in any investigation is the one written before it starts. It should record what is being investigated, expressed as specific allegations rather than general concerns; what is explicitly out of scope; who is commissioning it and to whom the investigator reports; whether the investigator is asked to make findings of fact only, or also to express a view on whether there is a case to answer; and an indicative timetable.

Without this, scope creep is invisible. With it, scope creep becomes a visible decision that somebody has to authorise — which is exactly what it should be. If new allegations emerge mid- investigation, that is not a reason to quietly widen the net; it is a reason to go back to the commissioner and formally amend the terms.

2. Decide the independence question honestly

An investigator needs to be independent in fact and to appear independent. In a small organisation, or where the respondent is senior, an internal investigator may be procedurally impeccable and still have their independence attacked — because they report to the respondent, work in the same team, or will be appraised by someone with an interest in the outcome.

Be honest about this at the outset. The question is not "can this person be fair" but "can this person's fairness be credibly challenged later". Where the answer is yes, an external investigator removes an entire line of attack for a cost that is usually modest relative to the risk. We provide this as a service — see employee relations and investigations — but the principle applies whoever you appoint.

3. Tell the respondent what they are accused of

Specifically, in writing, before the first meeting. Not "concerns have been raised about your conduct", but the actual allegations with enough particularity that a person could prepare a response. This is basic fairness, it is what any policy worth the name requires, and it is routinely done badly because the investigator is worried about tipping someone off.

Where there is a genuine risk of evidence being destroyed or witnesses being pressured, that is a reason to sequence the investigation carefully. It is not a reason to keep the respondent in the dark about the case they have to answer.

4. Take proper notes, and offer them for agreement

Investigation notes get disclosed. They get read by the respondent, by their representative, by an appeal panel and potentially by a tribunal. Notes that are fragmentary, editorialised, or written up from memory three weeks later are a liability.

The standard to aim for: a reasonably full record of what was asked and what was said, taken at the time, typed up promptly, and sent to the interviewee to check and confirm. If they disagree with the record, note the disagreement rather than arguing about it — a documented dispute about the note is far less damaging than an undocumented one.

Avoid writing anything in an investigation note that you would not want read aloud. Informal commentary, speculation about motive and personal characterisations are all disclosable and all unhelpful.

5. Interview the people who matter, not everyone

There is a tendency, particularly in nervous organisations, to interview everybody who might conceivably know something. This extends the timeline, spreads knowledge of the matter across the organisation, and produces volumes of marginal evidence that obscure the relevant material.

Work from the allegations. For each one, ask who has direct knowledge, what documentary evidence exists, and what would actually resolve the factual dispute. Interview those people. If the picture changes, interview more — but do it deliberately rather than by default.

6. Move fast, and explain any delay

Delay is the most common and most corrosive procedural failing. An investigation that takes six months causes real harm — to the respondent, who is often suspended or working under a cloud; to the complainant, whose confidence in the process erodes; and to the organisation, whose position becomes harder to defend the longer it takes.

The usual cause is not complexity but capacity: the investigation was assigned to a manager who already had a full job, and it competes with their day-to-day work every week. If you cannot release someone properly, that is an argument for appointing an external investigator or bringing in interim casework capacity — not for accepting a six-month timeline.

Where delay is genuinely unavoidable — a key witness on long-term sick leave, a parallel police investigation — document the reason and communicate it to the parties. Unexplained delay looks like indifference. Explained delay usually does not.

7. Keep investigation and decision separate

The investigator gathers evidence and makes findings of fact. The decision-maker decides the outcome. Collapsing these roles — where the investigator also chairs the hearing, or effectively dictates the sanction — undermines the whole structure and is straightforwardly unfair.

This is harder in small organisations, where the pool of available people is limited. It is still worth protecting. If you genuinely cannot find an internal decision-maker who has had no involvement, that is a strong argument for an external chair.

8. Write the report for a reader who was not there

A good investigation report sets out the terms of reference, the process followed, who was interviewed and when, the evidence considered, the findings of fact with the reasoning that supports each one, and — if the terms of reference ask for it — a view on whether there is a case to answer.

Write it as though the reader is a tribunal panel with no prior knowledge of your organisation. That discipline forces you to explain your reasoning rather than assert conclusions, and it tends to expose findings that are not actually supported by the evidence. Where the evidence is genuinely inconclusive, say so. An investigator who resolves every disputed point in a single direction invites the obvious question.

The underlying principle

Everything above reduces to one idea: run the investigation as though somebody will examine it closely, because in the cases that matter, somebody will. That is not defensive practice — it is simply what fairness looks like when it is written down.

If you are dealing with a matter where independence would be challenged, or where an investigation has already stalled, that is the point at which an external investigator is worth considering. We cover this under employee relations and investigations.

Note. This article is general guidance, not legal advice, and it does not create a client relationship. Employment law and statutory thresholds change; verify the current position before relying on anything here. For advice on your specific situation, get in touch.