How to conduct effective interviews in a workplace investigation

In most workplace investigations, the decisive evidence is not in a document: it is in what people saw, heard and did. That makes the interview the heart of the case file, and also its most fragile point. A poorly conducted interview contaminates testimony, tips off people who should not have been tipped off, and produces statements a court will discard without effort.

The good news is that interviewing well is not a mysterious talent: it is a method. These are the decisions that matter most.

Witness order is not a detail

The right sequence is almost always the same: the reporting person first, then peripheral witnesses (from least to most involved) and the respondent last. The reason is evidentiary: by the time the respondent speaks, the investigator knows the facts in enough detail to test their version, and the respondent can no longer adjust their account to what others said, because they do not know what others said.

Reversing this order (starting with the respondent "to hear their side") is one of the most frequent and most expensive mistakes: it buys time to align stories, pressure witnesses or destroy evidence.

Preparation: the interview is won before it starts

A serious investigator arrives with a preliminary chronology, the relevant documents read, and a topic guide (not a rigid script). They know what needs to be established and which elements of the conduct are in dispute. A protocolized guide, like the one used in the Wicklander-Zulawski method, orders the conversation into phases: rapport, free narrative, probing and closure.

Preparation also means deciding logistics deliberately: a private room, no interruptions, reasonable duration, and a conscious decision about who else is present. Every additional person in the room is an inhibiting factor.

Questions that build and questions that destroy

The central rule: open questions first, closed questions later, leading questions never. "Tell me what happened that day" produces information; "Is it true that he harassed her?" produces useless testimony, because the answer was planted by the investigator.

• Free narrative first. Let the person tell everything without interruption. Early interruptions cut off details you did not know you needed.
• Funnel later. From the general account to the specifics: dates, places, exact words, who else was present.
• Contrast at the end. Inconsistencies with other testimony or documents are raised at closure, neutrally and without accusation.

The record is the testimony

Testimony that was not properly documented barely exists, legally speaking. The minimum standard: an interview record with date, time, place, persons present, confidentiality and non-retaliation advisements, the account captured as faithfully as possible, and the interviewee’s signature. If the person corrects something before signing, even better: the correction proves they read and validated the content.

A note on recordings: they can help, but their legality and effects vary by jurisdiction and by consent. The decision to record is a legal one, not merely practical.

The mistakes that ruin the most case files

• Promising absolute confidentiality, or outcomes the investigator does not control.
• Group interviews "to save time": testimonies contaminate each other.
• Letting weeks pass between complaint and interviews: memory degrades and versions align.
• Having a party’s direct manager conduct the interview: impartiality is compromised from the start.

A well-conducted interview protects everyone: the reporter, whose account is faithfully recorded; the respondent, whose version is heard under the same rules; and the company, whose resulting file can stand up to anyone who reviews it.

← All insights