Reference checks are the only stage of hiring where you can ask someone who isn't trying to sell you the candidate. That's a rare advantage. Most teams waste it.

The classic question — "Would you hire them again?" — is useless. Almost every former manager will say yes. Most are answering politely; a few are answering legally. Either way, you get no information. The question is structured to elicit a yes.

Better questions ask for specific behavior under specific circumstances. "Tell me about a time they disagreed with you on a technical decision — how did they handle it?" The candidate's reference can't fabricate a story they don't have. The presence or absence of an answer is itself signal.

Five questions that actually work for us: (1) **"Where in your team would they rank against your current top performers?"** — forces a comparison. (2) **"What's the kind of work you wouldn't give them?"** — surfaces what they're not good at. (3) **"Walk me through a project that didn't go well — what was their role?"** — reveals failure response. (4) **"Who else on the team would you want me to talk to about them?"** — gets you an off-list reference. (5) **"If I told you they got the offer, would anything in their first 30 days surprise you?"** — pulls forward the concerns the reference wouldn't have volunteered.

Question 4 is the highest-leverage one. The candidate gave you a list of references they curated; the off-list name is one they didn't. People listed by their references are usually more honest, because they're a degree removed from the candidate's curation. We call the off-list name 80% of the time. They almost always pick up.

Treat vague answers as red flags, not as the reference being busy. "Yeah, they were great" with no specifics means the reference either didn't work closely with them or has nothing concrete to say. Press once with "Can you give me an example?" If still vague, ask question 4 to find someone who can.

Reference checks aren't a final ratification of a decision you've already made. They're a final chance to discover what the interview process missed. Run them like an investigation, not a rubber stamp.