Most reference checks are worthless.
Not unhelpful. Not marginal. Worthless. They consume time, generate paperwork, and produce exactly zero useful signal about whether a candidate will succeed or fail.
I say this confidently because I've seen the pattern hundreds of times. A hire looks perfect—strong interviews, great cultural fit, exactly the technical skills needed. References? They called three people, asked generic questions, got generic positive feedback. Standard process, checked box.
Three months later: PIP. The new hire couldn't work with others—territorial about his code, dismissive of feedback, unwilling to collaborate. He'd been fired from his previous role for exactly the same reasons, but none of the references had mentioned it.
When the hiring manager investigated afterward, the pattern was clear. The references provided were carefully chosen friends who would say positive things. The questions asked were generic and easy to answer favorably. Nobody had probed beneath the surface. The reference check had been ritual rather than investigation.
Reference checks are one of the most valuable tools in hiring—and one of the most consistently wasted. Most companies treat them as a compliance exercise: call the names provided, ask whether the candidate was employed, get some vague positive feedback, move on. This process generates almost no signal. The candidates who look good in interviews and provide friendly references will give good references; the candidates with serious problems will also give good references because they curated the list.
Getting real signal from references requires treating the process as investigation rather than validation. You're not confirming what you already believe; you're looking for information you don't have. The techniques matter.
At SmithSpektrum, we treat reference checks as a core part of evaluation, not an afterthought[^1]. Done well, references reveal things that no interview can surface—patterns of behavior over time, how someone acts when no one's watching, what happens when things get hard. Done poorly, they're waste of time.
Why Standard Reference Checks Fail
The typical reference check fails to generate signal for predictable reasons.
Candidates choose their references. They pick people who will say positive things—friends, mentors, allies. They don't list the colleague they clashed with, the manager they struggled under, or the peer who saw them at their worst. The sample is self-selected for favorable reviews.
Standard questions are easy to answer well. "What are John's strengths?" Every reference can provide strengths. "Would you hire him again?" People rarely say no directly. "What areas could he improve?" References have rehearsed responses—"he works too hard" or "he could delegate more"—that sound like weaknesses but aren't.
References don't want to give negative feedback. Even if they have concerns, they worry about legal liability, about the candidate finding out, about being the person who killed someone's opportunity. The path of least resistance is vaguely positive.
Phone dynamics favor positivity. A fifteen-minute call with a stranger doesn't create trust. References default to safe, positive statements rather than vulnerable, honest assessments.
Time constraints prevent depth. When you're checking three references before an offer deadline, each call gets rushed. There's no time to build rapport, follow threads, or probe beneath initial responses.
The result is uniformly positive feedback that doesn't differentiate strong candidates from problematic ones. The reference check box is checked, but no information has been gained.
Expanding Beyond Provided References
The first principle of effective reference checking is talking to people the candidate didn't provide.
Backchannel references—people you know or can reach who have worked with the candidate—provide unfiltered feedback. They weren't chosen by the candidate to say positive things. They have no incentive to present a curated view. Their perspective is more likely to be honest.
| Question Approach | What It Reveals | Red Flag Response | Green Flag Response |
|---|---|---|---|
| "Would you hire them again?" | Overall assessment | Hesitation, qualification | Immediate "absolutely" |
| "What would you coach them on?" | Development areas | "Nothing" (unrealistic) | Specific, minor items |
| "How did they handle conflict?" | Collaboration style | Avoidance or blame | Constructive resolution |
| "What was their biggest impact?" | Scope and contribution | Vague or small | Specific, significant |
| "Who else should I talk to?" | Network and relationships | Can't suggest anyone | Multiple names readily |
Finding backchannel references requires effort. Look at the candidate's LinkedIn. Who do you know who overlaps with their experience? Who do your colleagues know? Who can reach someone who worked with them? The tech industry is interconnected; two degrees of separation usually reaches someone.
When requesting backchannel references, be clear about confidentiality. The reference should understand that you won't reveal their feedback to the candidate. This creates safety for honest answers.
The combination of provided references and backchannel references gives you multiple perspectives that can be compared. If provided references say one thing and backchannel references say another, that's significant information.
Questions That Actually Work
The questions you ask determine the signal you get.
Start with rapport building before you probe. A minute or two of conversation—how they know the candidate, what context they worked in together—builds enough relationship for more honest answers. Jumping straight to questions feels like interrogation and produces defensive responses.
Ask for specific examples, not general assessments. "Can you tell me about a time when Alex dealt with conflict on the team?" gets more information than "How does Alex handle conflict?" Stories reveal behavior; generalities can be shaped to sound good.
Use scaled questions to force differentiation. "On a scale of 1 to 10, where would you rank Alex among engineers you've worked with?" If the answer is 7, you can ask what would make them a 10. If the answer is 10, you can ask what distinguished them from an 8. Scales force references beyond generic positivity.
Ask about specific concerns you have. If you're wondering about collaboration, ask directly: "In your experience, how did Alex work with people who disagreed with him?" If you're wondering about technical depth, ask: "Were there technical areas where Alex needed more support than others?"
Ask what the candidate struggled with. Not "areas for improvement" (which sounds like a performance review)—"struggles." What was hard for them? Where did they have friction? What didn't go well? This question often yields more honest answers than asking about weaknesses.
Ask about trajectory. "Did Alex improve over the time you worked together? In what ways?" This reveals learning ability and self-awareness—whether they recognized and addressed their gaps.
Ask the question that matters most: "Would you enthusiastically hire them again for this specific type of role?" Not "would you hire them"—people say yes to that too easily. "Enthusiastically" creates space for honest hesitation. Probing the hesitation often reveals important information.
Reading Between the Lines
What references don't say often matters as much as what they do say.
Faint praise is a signal. "He was competent" from someone who worked closely with the candidate is not a strong endorsement. If you'd expect enthusiasm and get lukewarm approval, something is being left unsaid.
Hesitation before answering suggests internal conflict. The reference knows something they're deciding whether to share. A question like "You paused there—is there something on your mind?" sometimes opens up honesty.
Qualified answers indicate complexity. "In certain situations, she was excellent" suggests that in other situations, she wasn't. Ask about the other situations.
Narrow praise can indicate narrow strengths. If every positive thing the reference says is about technical skills, maybe soft skills are weak. If everything is about work ethic, maybe output quality was lacking.
Changes in tone or energy matter. If a reference is enthusiastic about some topics and flat about others, the flat topics may be concerns.
None of these signals are conclusive—people speak differently for many reasons. But patterns across multiple references are meaningful. If three references all hesitate when asked about collaboration, collaboration is probably a real issue.
Questions for Different Concerns
Different potential concerns require different questions.
For collaboration concerns: "Can you describe how they worked with people who had different views?" "What happened when there was conflict on the team?" "How did they respond to feedback they disagreed with?"
For autonomy concerns: "How much direction did they need?" "Were there times when they got stuck and needed help? What did that look like?" "Could they scope their own work, or did someone need to define it for them?"
For quality concerns: "What was the quality of their work typically like?" "Were there patterns in the mistakes they made?" "How did they respond when bugs were found in their code?"
For reliability concerns: "Did they meet their commitments?" "Were there times when projects didn't go as planned? What happened?" "How did they handle deadlines?"
For leadership concerns: "Did other engineers seek them out for help or guidance?" "What influence did they have on the team beyond their own work?" "How did they handle situations where they needed to lead without authority?"
For cultural fit: "What kind of environment brought out their best work?" "Were there aspects of your workplace culture that didn't work well for them?" "What type of manager or team would you recommend for them?"
When References Conflict
Sometimes references disagree—one person is enthusiastic, another is lukewarm.
First, consider context. A reference who worked closely with the candidate daily has more signal than one who collaborated occasionally. A recent reference is more relevant than one from five years ago. A reference in a similar role provides more applicable insight than one from a different context.
Different experiences might reflect real variation. Some people perform well with certain managers or in certain environments and poorly in others. If one reference worked well with the candidate and another didn't, understanding why helps you assess whether your environment will be a good or bad fit.
Conflicting references often reveal something important about the candidate. Maybe they're excellent with some personalities and difficult with others. Maybe they thrive in some contexts and struggle in others. The conflict is information.
If concerns from one reference are serious, probe deeper. Ask additional references about the specific concern. Talk to more people if needed. A single negative reference might be an outlier or a grudge; multiple references with the same concern is a pattern.
Reference Check Timing
When you do reference checks affects what you learn and how you use it.
Late-stage reference checks (after offer decision, before extending) are the norm but have limitations. By this point, you've decided you want to hire. You have confirmation bias—you want to confirm the decision you've already made. Negative information feels inconvenient rather than valuable.
Earlier reference checks can be more useful. If you check references before final interviews, you can probe concerns during those interviews. If you check before the offer decision, the information actually influences the decision.
The tradeoff is efficiency. Reference checks take time, and you don't want to invest that time on candidates who won't make it to offer. But checking only at the end means information rarely changes outcomes.
A middle path is doing lightweight backchannel references early (quick pings to your network about candidates you're considering) and thorough reference checks before final decisions. This balances efficiency with genuine information-gathering.
When Reference Checks Reveal Problems
Sometimes reference checks surface serious concerns late in the process. What then?
Don't ignore the information because you like the candidate. If references reveal a pattern of concerning behavior, that pattern is likely to continue. The interview shows you the candidate at their best; references show you the candidate in ongoing work situations.
Probe further before making decisions. One lukewarm reference might be noise; multiple concerns are a pattern. If serious issues surface, talk to more people. Don't decide based on incomplete information.
Consider whether the concern is disqualifying or addressable. Some issues are fundamental—dishonesty, repeated performance failures, inability to work with anyone. Others are context-dependent—someone who struggled in one environment might thrive in yours. Others are developmental—a growth area that you can support.
Be honest with yourself about what you can address. "We'll coach them through it" is often wishful thinking. Some issues are coachable; many aren't. If the reference check reveals patterns that would be problems in your environment, believe the pattern.
The team that hired the brilliant-but-impossible engineer? They overhauled their reference process.
They started checking backchannel references—reaching out to people in their network who'd worked with candidates. They asked harder questions: specific examples of conflict, struggles the candidate had experienced, whether they'd be enthusiastically re-hired for this specific type of role.
The next hiring round, a candidate interviewed brilliantly but backchannel references told a different story. Technically excellent but condescending to junior engineers. Previous team had been relieved when he left. Not openly difficult—but subtly undermining.
They passed on the candidate. He was hired elsewhere and, they learned later, left under similar circumstances within a year.
"References aren't about checking boxes," the hiring manager reflected. "They're about finding out what the interview can't tell you. But you have to actually try to find out."
References
[^1]: SmithSpektrum talent advisory, reference check methodology, 2019-2026. [^2]: Geoff Smart and Randy Street, "Who: The A Method for Hiring," 2008. [^3]: Google re:Work, "Reference Check Guidance." [^4]: SHRM, "Effective Reference Checking," 2024.
Building an effective hiring process? Contact SmithSpektrum for interview design and evaluation methodology.
Author: Irvan Smith, Founder & Managing Director at SmithSpektrum