Two Complaints and One Referral. Who Sets the Scope When You Hire an Investigator

A fourteen-person civil engineering firm in the Midwest received two written complaints about the same project lead inside a single month, one about a comment made in a client meeting and one about scheduling retaliation. The owner, who also stamps drawings and chases receivables, asked a peer at a trade association who they used, got a name, looked at a website with a row of certification acronyms across the footer, and hired the consultant by email. The email ran two sentences. It said the firm needed someone to look into concerns raised about a manager and asked how soon they could start.

Those two sentences did more work than the certifications did. They defined the scope, the deliverable, the audience for the findings and the standard the work would be judged against, all by leaving each of them blank. That is the part almost nobody traces back later, when the consequences arrive.

What the engagement email actually decided

The consultant arrived, interviewed six people over two days, and delivered a nine-page memo that summarized what each person said, noted where accounts diverged, and recommended coaching. It did not state whether the conduct described violated the firm's own handbook, because nobody had asked it to, and it did not reach a conclusion on the retaliation complaint, because retaliation involves a timeline and the consultant had not been given calendars, assignment logs or emails. The memo was competent. It was also, for the two purposes the firm would eventually need it for, unusable. The gap was created in the engagement, not in the fieldwork.

An investigation produces one of three things: a narrative, a finding, or a finding measured against a written standard. They cost different amounts, take different lengths of time, and survive different levels of later scrutiny. A buyer who does not name which one they are purchasing will receive whichever one the provider habitually produces, and most providers habitually produce the cheapest defensible version, which is the narrative.

The party being addressed is not the party who decides

In this transaction the consultant was pitching the owner, and the owner believed he was the decision maker, because he signed the check. He was not, in any sense that mattered six months out. The people who would eventually decide whether the work was adequate were a state unemployment hearing officer, an attorney representing the project lead after his termination, and a claims examiner at the firm's employment practices carrier, each of whom would read the same nine pages looking for something it did not contain. The owner was buying a document for readers he had never met and had not thought to ask about.

This reversal shows up across professional services, and it is the most reliable single test of whether an engagement is well specified. Ask who reads the output, in what setting, and with what authority to reject it. A structural letter is read by a plan reviewer. A valuation is read by an examiner or an opposing expert. An investigation report is read by whoever is deciding a claim. Naming that reader forces the provider to describe the standard they work to, and that description is where good separates from plausible.

Four questions that a plausible provider answers differently

The firm's second attempt, eight months later and after a demand letter, was scoped in a call that lasted under an hour. The owner asked what the deliverable would be and got a direct answer: findings on each allegation, stated as substantiated, not substantiated, or unable to determine, measured against the handbook language in effect on the date of the conduct. He asked what documents the investigator required before starting, and got a list that included the handbook, prior discipline records, the assignment scheduling system and the relevant email accounts, with a note that the engagement would be paused rather than completed if access was refused. Plausible providers say they will work with whatever is available.

He asked who the investigator would report to, and whether the report should come through counsel. The answer distinguished the two routes plainly: engaged directly, the report is a business record that the firm can hand to its carrier and rely on, and engaged through counsel for the purpose of legal advice, it may be protected but the firm gives up the ability to wave it around freely. He asked, last, what the investigator would not do, and heard that they would not make the termination decision or recommend one, because an investigator who recommends the outcome has merged two roles that a hearing officer will later try to pull apart. That refusal is a mark of the real thing.

The consequences people rarely trace back

The costs of the first engagement did not appear on the invoice. They appeared as a second investigation covering the same events with witnesses whose memories had aged eight months, as an unemployment determination the firm lost because it could not show a finding of misconduct, as a carrier reservation of rights that turned on the firm's failure to document that it had investigated to a standard, and as three engineers who watched nothing happen after two written complaints and quietly updated their resumes. None of those five outcomes reads, at the moment it lands, like a consequence of a two-sentence email. All five are.

The federal agency responsible for enforcing workplace discrimination law, the Equal Employment Opportunity Commission, oversees this area, and its existence is precisely why the audience question matters: an employer's response to a complaint is itself evaluated, separately from the underlying conduct. A firm that scopes the work to produce findings against a written standard, with a defined document set and a named reader, has bought something that holds. A firm that buys a summary has bought reading material.

The useful habit is small and unglamorous. Before hiring any outside expert, write down who will read the output and what they are empowered to reject, then send that paragraph to the provider and see whether their proposal changes shape in response. The ones who ask three follow-up questions and revise the deliverable are worth the fee. The ones who confirm availability are selling you the thing they already make.