Priya is forty minutes into her Tuesday, on the third reference call of the morning, when something stops making sense. The candidate, Marcus, listed his reference as a former teammate, someone junior to him on the same project, the kind of reference most hiring managers treat as a formality: friendly, unlikely to complicate things, probably going to say something warm and vague so everyone can move on. Except the man on the phone isn't talking like a peer. He's talking like a manager. He describes Marcus's "development areas" in the structured, faintly HR-inflected language of a performance review, not the language of someone who sat two desks away. Priya asks a direct comparative question, how Marcus's output stacked up against two other named people on that team, and the answer comes back thin and generic, the kind of thing you'd say about someone you'd heard of rather than worked next to. She has the candidate's CV open in one tab and a shared reference-check script in the other, the kind of call that usually takes fifteen minutes and ends with everyone moving on to the next name on the list. This one doesn't.
The assumption most hiring managers carry into reference calls is that if a reference is going to be faked, it'll be faked easy: a friend, a junior colleague doing a favor, someone with nothing to lose and no real standing to contradict the story. The data doesn't fully support that picture. A survey by Standout CV found that roughly one in six respondents admitted to having faked a reference at some point in their career, already a higher number than most hiring teams would guess. What's more surprising is who. In that same data, around 40% of the fake references came from applicants who were themselves at mid-management level, not fresh graduates smoothing over a thin work history. The people most likely to construct a fake reference are often the people experienced enough to know exactly how to make one convincing.
That tracks with what happens on calls like Priya's. A junior contact doing a favor for a friend is easy to catch, because they usually haven't rehearsed anything and will say so if pushed. A more senior, more deliberate fake, arranged by someone who understands how a reference call is supposed to sound, is a different problem. It's built to survive an open-ended "tell me about them," precisely the kind of question most reference checks still lead with.
A well-prepared fake reference usually gets the easy parts right: the company name, the rough dates, a job title that matches the CV. What it struggles to fake convincingly is specificity under pressure, the granular, comparative detail that only comes from having actually sat in the same meetings as someone, watched them ship something, disagreed with them about a decision. Marcus's reference had clearly rehearsed the broad strokes. He hadn't rehearsed for a question that assumed real, textured familiarity with two other named colleagues, and it showed within about ten seconds.
The scale of this isn't a rounding error, either. A survey by the BBC and the Federation of Small Businesses, covering more than 1,000 companies, found that 17% had discovered, after the fact, that an employee had faked at least one reference during hiring. Across employers more broadly, 29% reported having caught a fake reference at some point in their history. For a meaningful share of hiring teams, this has already happened. Most of them just don't bring it up at the next hiring meeting.
What actually caught Marcus's reference wasn't instinct, it was structure. Priya's comparative question, how does this person's output stack up against two specific named colleagues, is hard to fake convincingly because it demands detail a genuine peer would have and a rehearsed one usually doesn't: specific comparisons, specific projects, the texture of having actually sat in the room. Open-ended prompts invite a prepared answer. Comparative, specific ones expose the difference between someone who worked alongside a person and someone who was told what to say.
None of this means every reference is suspect, or that most candidates are lying about who worked with them. The point is narrower: the reference most likely to be fabricated is not the one that looks easiest to fake. It's the one nobody thought to double check, because it looked ordinary. Marcus's reference looked ordinary right up until it didn't.