Sexual Harassment Investigation Procedures: Step-by-Step

6/14/2026

Sexual harassment investigations differ from other workplace investigations in ways that matter procedurally, not just emotionally. The conduct is usually unwitnessed, the evidence is usually testimonial, the complainant is frequently experiencing a stress response that affects how they recount events, and the legal framework asks questions — welcomeness, severity, pervasiveness — that other investigations do not.

Quick answer: Handle sexual harassment matters with the standard investigation process plus four adaptations: trauma-informed interviewing, careful evidence gathering aimed at corroboration rather than eyewitnesses, an analysis that addresses welcomeness and severity or pervasiveness without putting the complainant on trial, and heightened attention to interim measures and retaliation.

The Legal Framework the Investigation Must Address

Your report finds facts and policy violations, not legal conclusions. But you need to know what facts matter.

Element	Facts to Develop
The conduct occurred	Each incident: date, location, who was present, what was said or done, what happened immediately after
It was unwelcome	How the complainant responded; whether they objected, avoided, or reported at the time; contemporaneous statements to others
Severity or pervasiveness	Frequency, duration, whether physical, whether threatening or humiliating, whether it interfered with work
Employer knowledge	Whether and when anyone in management knew; prior complaints about the same respondent
Employer response	What was done, when, and whether it was effective

A caution on welcomeness. This element is legitimate to investigate and easy to investigate badly. Focus on how the complainant responded to the conduct — not on their clothing, social life, relationship history, or prior consensual conduct with others. Questions in the latter category are irrelevant, damaging, and in many jurisdictions restricted.

Trauma-Informed Interviewing

Stress and trauma affect memory encoding and recall in ways that look like inconsistency to an untrained interviewer. Understanding this changes both how you interview and how you assess credibility.

What trauma responses look like

• Fragmented chronology. Central sensory details may be vivid while sequence and peripheral details are unclear.
• Flat affect. Absence of visible distress is not evidence that nothing happened.
• Additional recall over time. New details emerging in a second interview are common and not inherently suspicious.
• Delayed reporting. Extremely common, for reasons including fear of retaliation, self-blame, and uncertainty.
• Continued interaction with the respondent. Employees frequently maintain apparent normalcy afterward, particularly with a supervisor.

Interview adaptations

1. Give the complainant control. Let them choose the location and timing where possible, and offer a support person where policy permits.
2. Explain the process first, including what you can and cannot promise.
3. Start with an open narrative and do not interrupt. Let silences run.
4. Ask about sensory and experiential detail rather than chronology first: "What do you remember about that moment?" often produces more than "What happened next?"
5. Avoid "why" questions. "Why didn't you report it sooner?" reads as blame. Use "Help me understand the timeline between the incident and when you came forward."
6. Offer breaks and be willing to continue another day.
7. Do not require a chronological account in the first pass. Build the timeline afterward from what they provided.

Evidence: Corroboration, Not Eyewitnesses

Sexual harassment is rarely witnessed. Build the case from indirect corroboration:

• Contemporaneous statements. Who did the complainant tell at the time? A text to a friend on the day of the incident is powerful corroboration precisely because it predates any complaint.
• Message and email records between the parties, including tone and frequency changes.
• Timing and access evidence. Badge records, calendars, and schedules establishing opportunity.
• Behavioral change evidence. Attendance, performance, schedule requests, or transfer requests around the relevant period.
• Prior complaints about the same respondent — check whether the organization has records, including informal ones.
• Pattern witnesses. Other employees who experienced or observed similar conduct. Ask every witness whether they have observed similar conduct toward anyone.
• Physical evidence where applicable — photographs, gifts, notes.

Interim Measures

This is where employers most commonly create a retaliation claim while trying to help.

The rule: the burden falls on the respondent, not the complainant.

Avoid	Instead
Moving the complainant's desk or shift	Move the respondent, or adjust their schedule
Transferring the complainant to another team	Change the respondent's reporting line or work location
Placing the complainant on leave	Suspend the respondent with pay if separation is needed
Removing the complainant from a project	Remove the respondent's supervisory authority over the complainant

If the complainant requests a change, document that it was their request and confirm in writing.

Interviewing the Respondent

Fairness requires that the respondent understand the allegations well enough to respond meaningfully. A vague summary produces a vague denial and an unfair process.

• Present each allegation specifically — date, location, conduct
• Ask for their account of each
• Ask about each piece of contrary evidence
• Ask whether they have witnesses or documents
• Ask directly whether they have any reason to believe the complainant would fabricate
• Ask whether similar allegations have been made before
• Advise of the anti-retaliation obligation in writing

Do not disclose the complainant's identity where it can be avoided — though specific allegations usually make it apparent. Weigh fairness against retaliation risk and document the reasoning.

Credibility Without Stereotypes

Assess credibility against articulated factors: consistency across accounts, level of detail, contemporaneous corroboration, plausibility, motive to falsify, and any prior record.

Explicitly do not weigh:

• Delay in reporting, standing alone
• Continued interaction with the respondent
• Absence of visible emotion
• Imperfect chronology
• The complainant's demeanor measured against an assumed "typical" reaction
• The respondent's reputation or seniority

Where you consider a factor that could cut against the complainant, address it explicitly in the report and explain your reasoning. Acknowledging and analyzing the weakness is what makes the conclusion credible.

Closure and Follow-Up

• Inform both parties in writing that the investigation concluded.
• Tell the complainant that appropriate action was taken where warranted. You generally should not disclose the specific discipline imposed on another employee.
• Reiterate anti-retaliation protection to both parties, in writing.
• Check in with the complainant at 30 and 90 days to confirm the conduct stopped and no retaliation occurred. Document each check.
• Consider whether the matter indicates a broader problem requiring training, policy change, or a climate assessment.

Frequently Asked Questions

What if the complainant asks us not to investigate?

Once on notice, the employer generally must act. Explain why, discuss steps to reduce their exposure, and document the conversation.

Does delayed reporting undermine credibility?

Not on its own. Delay is common and has well-documented causes. Explore the reason neutrally and weigh it alongside other factors.

Can we ask about the complainant's relationship history?

Generally no. Prior consensual conduct with others is not relevant to whether this conduct was unwelcome, and such questions are restricted in many jurisdictions.

What if the conduct is not substantiated?

"Not substantiated" is not "did not happen." Corrective measures may still be appropriate — policy reminders, training, or separating the parties without characterizing it as discipline.

Should we use an external investigator?

Consider it when the respondent is senior, when HR is implicated, when there are multiple complainants, or when litigation is likely.

These Investigations Reward Training Most

Sexual harassment investigations are where untrained investigators do the most damage — to the complainant, to the respondent's due process, and to the organization's defensibility.

The Internal Investigations Certificate Program covers interviewing, credibility assessment, and report writing. The Advanced Internal Investigations Certificate Program goes deeper on sensitive and high-exposure matters.

👉 See the Workplace Investigation Training Program →

Additional resources: How to Conduct a Harassment Investigation | Guide to Investigating Harassment Complaints | EEOC harassment guidance