Most companies treat reference checks as a formality: two calls, generic questions, done the week before an offer to confirm a decision that's already been made. That treatment made a certain amount of sense when résumés and interview performance were themselves reasonably hard to fake. They aren't anymore. As every other signal in the hiring process gets noisier, easier to polish, easier to prepare for, the one signal that's still expensive to fabricate, an honest account from someone who actually worked alongside the candidate, has become relatively more valuable, not less. Most companies haven't updated how they use it.
Why reference checks are underused and badly done
The typical reference check happens at the very end of a process, after the hiring manager has already mentally committed, with two references the candidate hand-picked because they know exactly what those people will say. The questions asked are almost designed to produce nothing: 'would you recommend this person,' 'what are their strengths,' 'is there anything I should know.' Nobody answers 'no' to the first question even about a genuinely difficult employee, because declining to recommend someone feels like an aggressive act most people avoid. The result is a step everyone performs and nobody actually uses, checked off rather than genuinely relied on.
Why this signal has gotten relatively more valuable
Every other input into a hiring decision has gotten easier to polish. A résumé can be optimized in ninety seconds. Interview answers, including behavioral ones, can be rehearsed against widely available question banks, and increasingly practiced with AI tools that coach candidates through likely questions in advance. None of that makes candidates worse, plenty of strong candidates use these tools too, but it does mean the gap between 'performs well in a process' and 'is actually strong at the job' has widened. A reference check is one of the few remaining signals that's expensive to fabricate: it requires a real person who actually worked with the candidate to describe what actually happened, in specific terms, on the record.
What a good reference call actually asks
The difference between a useless reference check and a genuinely informative one is almost entirely in the questions. Generic questions produce generic, safe answers. Specific, behavioral, outcome-anchored questions produce specifics, because they ask the reference to recall an actual event rather than render a general verdict.
- "Tell me about a specific time this person disagreed with a decision. What did they do, and how did it get resolved?"
- "Describe a project that didn't go well while they were involved. What was their role in it, and how did they respond?"
- "If you were building a team again tomorrow, would you actively try to bring this person with you? Why or why not, specifically?"
- "What's something this person would say themselves that they need to work on? Would you agree?"
- "On a scale, how does this person's actual output compare to the strongest person you've managed in a similar role, and what's the gap?"
Call the people who managed the hard moments, not just the fans
Candidate-provided references are, reasonably, chosen because they'll say something positive. That doesn't make them worthless, but it does mean the most useful reference is often not the most enthusiastic one, it's the one who managed the candidate through something difficult: a missed deadline, a reorg, a project that got cut, a period of underperformance that was later corrected. Ask the candidate directly for a reference from a harder stretch, not only their best moment, and most strong candidates will provide one without hesitation, because they know it reflects well on how they handled adversity, not just how they perform when things are easy.
Timing: run references alongside the process, not after it
Reference checks scheduled after the interview loop is functionally over arrive too late to change anything, they become confirmation of a decision rather than an input into it, and it shows: hiring managers rarely act on something a late-stage reference raises, because reversing course after an internal decision feels costly. Run structured reference calls in parallel with your second or third interview round instead. Done this way, they either surface something worth probing directly with the candidate before an offer, or they add real confidence to a decision that's still genuinely open.
| Element | Treated as formality | Treated as real signal |
|---|---|---|
| Timing | After the loop, before the offer letter goes out | In parallel with round two or three, while the decision is still open |
| Questions | "Would you recommend them?" | Specific behavioral and outcome questions about real events |
| Who's called | Whoever the candidate lists first | Includes someone who managed the candidate through a hard stretch |
| What happens with the answer | Filed, rarely changes the outcome | Actively shapes the final decision or what gets probed with the candidate |
