Candidates are advised to supply a suitable list of referees, which tells you most of what you need to know about a candidate-supplied list.
Why it matters when the plan changes
References are the only stage of selection where someone who has actually seen the person work is asked about it, which makes them potentially the most informative evidence available among the signals a process collects, alongside interviews and role data. They are usually run as a formality at the end, with referees the candidate chose and questions that invite confirmation, which converts that potential into a ritual rather than a test.
The tension is between depth and fairness. The most informative conversations are with people the candidate did not nominate, which raises real questions about consent, confidentiality and a candidate's ability to respond to what is said about them. Where evidence is genuinely absent, a careful process marks that gap rather than filling it with the confidence a friendly referee supplies. The informal version of this happens constantly in senior hiring and is rarely acknowledged or recorded.
In practice
Three referees supplied by a candidate confirm the panel's impression. A fourth conversation, with someone who reported to the candidate at a previous employer, describes a pattern under pressure that nobody asked about because the questions were about strengths. Both accounts are honest. Only one of them was designed to be able to disconfirm anything.
Evidence
Candidates are advised to supply a suitable list of referees to a prospective employer, which shapes what a reference conversation can find.
Employment reference letter, Wikipedia (2026)People search for, favour and recall information that confirms what they already believe, which is what an end-stage reference call invites.
Confirmation bias, Wikipedia (2026)
What it cannot tell you
Reference checks reveal only what referees choose to say and only what they were asked to reflect on, not necessarily how the candidate will behave in the specific pressures of a new role. A conversation with a curated list, however honest, cannot substitute for observation of conduct under conditions the referee never witnessed.
Questions
Because the candidate selected them, reasonably, to be favourable. A 2026 Wikipedia entry on employment reference letters notes that candidates are advised to supply a suitable list of referees, which shapes what any conversation can find. The list can still inform if questions are specific and behavioural, but it is not a test.
Specific questions about observed behaviour under pressure, asked of people who saw the candidate struggle rather than at their best. Confirmation bias, as described in the 2026 Wikipedia entry, is the tendency to search for and recall information that confirms existing belief, which end-stage questions about strengths readily invite.
They happen constantly in senior hiring and raise genuine issues. A candidate cannot respond to something they were never told was said, and the information has no provenance. Where they are used, being open that the market will be asked is fairer than pretending otherwise.
Early enough that what they surface can still change the decision. Taken after an offer has effectively been made, they function only as confirmation, and a panel that has already committed will discount a discordant reference rather than reopen the question.
Who was asked, what they were asked, what they said including anything that argued against the appointment, and how it was weighed. A reference process with no record of disconfirming evidence cannot be distinguished afterwards from one that never sought any.