The behavioural reference: what former employers will not say outright

Reference checks used to be where you found out who a candidate really was. The previous manager would tell you, candidly, that the candidate was technically strong but difficult with peers, or that they had been managed out under a euphemism. That world is gone. Modern reference checks operate under legal-caution discipline: former employers, especially in regulated industries and large enterprises, are coached to disclose nothing that could expose the company to a defamation claim. The reference call has been hollowed out from a candid source to a verification of dates and titles.

This is not a story about decline. It is a structural shift in how information moves between employers. The information about conduct, ethics, and behaviour still exists; previous employers still know it. What has changed is the structure required to extract it. Generic questions get generic answers. Specific, structured, jurisdictionally-anchored questions get more useful signal, including the signal of evasion. The reference check is not dead. It has just stopped working for the people who do not know how to ask.

What standard reference calls actually capture

A typical standard package reference call covers a predictable set of questions: dates of employment, role title, scope of responsibility, performance during tenure, reason for departure, eligibility for rehire, and an open invitation to add anything else relevant. The form is well-intentioned but it produces a narrow band of signal.

"Would you rehire?" is the most common terminating question. The expected positive answer is a near-binary yes, which means almost nothing because references are coached to give it. A "no" or qualified "maybe" is rare and almost always means something significant. The middle ground, where most useful information sits, is not surfaced by this question at all. Eligibility for rehire is a company policy decision in many large employers, not a reflection of the individual; a former employer can legitimately answer "no" because of a no-rehire policy that applies to every departure, regardless of conduct.

Performance signal is similarly narrow. Most references will confirm that the candidate met expectations or exceeded them; very few will volunteer that the candidate was a low performer, both because of legal exposure and because most candidates select references who liked them. Performance information from a standard reference is essentially candidate-curated.

The structural problem is that the standard reference call is designed to confirm what the candidate has already told the hiring company. It is verification, not discovery. The conduct, ethics, and behavioural patterns that actually predict post-hire risk live outside this scope.

The structured behavioural framework

A structured behavioural reference is a reference call designed around a pre-defined set of conduct-specific questions, asked the same way to every reference, with the answers coded against an expected pattern. The framework is not improvised by the analyst running the call; it is a programme decision. The questions are deliberately phrased to allow the reference to answer with yes / no / prefer not to comment, which provides legal cover for the reference while still extracting useful signal for the hiring organisation.

The core question set, calibrated to a senior or people-leader hire:

Reading the signals

The structured framework produces three types of answer for each question: an answer, a non-answer, or a hesitation. Each is information.

The interpretation is not done on a single reference. The pattern is the product. Three references who all answer "prefer not to comment" to question 01 is a high-confidence signal that something exists. Three references who all answer "no, never" is the converse. The structured framework allows the pattern to emerge because the same question is asked the same way each time. A free-form reference call does not produce comparable answers and cannot pattern-match.

Operator-led vs form-completed references

The largest single failure mode in modern reference work is the form-completed reference. A junior analyst emails a questionnaire to the named reference, receives a partially-completed response sometime later, files the response into the BGV report, and the case closes. The reference never spoke. The framework was not executed. No signal beyond what was written down was captured.

Form-completed references fail in three ways: the reference can answer with maximum hedging because there is no live conversation; the verbal cues (hesitation, qualification, tone) are absent; and the candidate may have been involved in drafting the form responses, which is more common than vendors acknowledge. The hiring organisation receives a reference report that looks complete and is operationally hollow.

Operator-led reference work means a trained analyst calls the reference, executes the structured framework live, captures verbatim quotes and verbal cues, and codes the answers against the expected pattern. The analyst is the instrument. The protocol is the same across calls; the human reading of the answers is what makes the protocol work. This is the same operator-led principle that applies to other parts of source-level verification: depth comes from human judgment, not from form automation.

For PE operating partners: are your portfolio companies running real references?

The single highest-leverage question to ask a portfolio company CHRO is: walk me through how you do reference checks for a senior hire. Listen for three things. First, are the references conducted as live calls or as emailed forms? Second, are the questions structured and pre-defined, or open-ended? Third, when a "prefer not to comment" answer comes back, is it recorded and acted on, or filed as "no concerns surfaced"?

If the answer is "we send a form, we get whatever they fill in, we file it," the portfolio company is not doing reference work. They are completing paperwork. The harassment hire and the conduct-pattern hire will both pass this process. The next incident is structurally inevitable. The remediation cost will dwarf the incremental investment in operator-led reference work.

The portfolio-level standard worth pursuing: every senior or people-leader hire across the portfolio runs structured behavioural references, executed by trained analysts, with the answers coded and patterns matched. This is achievable with the existing BGV vendor relationship if the programme scope is rewritten. It is not achievable if the contract is for "standard reference checks" and the vendor is form-completing for cost reasons.