Are reference checks still worth it in 2026

HireRight’s tenth annual employment screening benchmark survey found that 85% of employers had caught a lie on a candidate’s resume or application, up from 66% just five years earlier. Candidates polish, exaggerate, and sometimes outright invent, and employers keep catching more of it, not less.

Meanwhile, talk to enough small business owners and you’ll hear the same quiet admission: a lot of them have stopped checking references altogether, and most of the rest are calling someone the candidate personally selected to vouch for them. Neither group is catching much.

That’s the part most hiring advice skips. Reference checks aren’t fading because owners got lazy. They’re fading because enough people who’ve run hundreds of them noticed the same thing. Nobody hands over the name of someone who’ll say something bad about them.

Reference checks feel like due diligence

Search “are reference checks worth it” and you’ll get a consistent answer: yes, but do them right. Ask standardized questions across a handful of categories. Get at least two or three references, not one. Time the calls after you’ve made a conditional offer, not before. Ask “would you rehire this person,” because it’s supposedly the single most predictive question you can ask.

All of that is reasonable advice for running a reference check well. None of it touches the actual problem.

You can standardize the questions, train every manager on your team to ask them the same way, and still end up with the same three names on every form: a manager who liked the candidate, a peer who liked the candidate, and a mentor who liked the candidate. The process looks rigorous. The input never changes.

This is why reference checks became a formality that almost everyone still performs and few people actually trust. Talk to enough owners who hire the same roles over and over and you’ll hear the same admission, usually said half-jokingly: they do it because skipping it feels riskier than doing it, not because the call ever tells them something they didn’t already suspect.

The candidate builds the witness list

Here’s the structural problem, and it’s the same one that’s been quietly breaking resumes for the past few years. A resume is written by the candidate, about the candidate. A reference is chosen by the candidate, to talk about the candidate. Both are curated. Only one of them pretends not to be.

Nobody puts a mediocre former boss on their reference list. Nobody includes the coworker they clashed with. The list is a selection of the three people, out of everyone the candidate has ever worked with, most likely to say something flattering. That’s not a flaw in how any individual owner runs the call. It’s the design of the exercise.

The data backs this up further back than the current conversation about it. In Schmidt and Hunter’s widely cited 1998 meta-analysis of personnel selection methods, reference checks scored a validity coefficient of .26 for predicting job performance. That’s lower than structured interviews, work sample tests, and even unstructured interviews. Among the methods employers commonly rely on, only raw years of job experience and years of education scored worse. Reference checks were never in serious contention to begin with, because the exercise was never built to measure performance. It was built to confirm that someone, somewhere, would say a nice thing about the candidate on the record.

For an owner hiring the same handful of roles over and over, this isn’t an abstract methodology critique. It’s real, repeated time spent on calls that confirm what you already assumed and rarely change a decision. On a four or five person team, the hire you’re worried about getting wrong is exactly the one a curated reference will never flag.

To be fair, reference checks still catch some things. They confirm dates, titles, and eligibility for rehire, which matters for compliance and for weeding out outright fabrication. That’s worth keeping. What they were never going to tell you is how someone actually performs, communicates, or handles pressure, because the person answering the phone was chosen for the answer they’d give.

Gather the harder-to-fake evidence first

Once you see the reference check for what it is, a curated endorsement rather than an observed behavior, the fix isn’t a better call. It’s moving the evidence-gathering earlier, to a point in your process the candidate can’t hand-pick.

A reference is a report from someone the candidate chose. A structured interview response or a work-sample assessment is a record of what the candidate actually said or did, on the record, with no friendly witness standing between you and the behavior. That’s the one thing worth taking from this: stop asking a chosen witness to describe the candidate’s character, and start collecting evidence of how the candidate actually shows up, earlier in the process, before there’s anyone left to curate.

This is where Truffle fits into the picture. Truffle is a candidate screening platform that combines resume screening, one-way video interviews, and talent assessments, so the evidence you’d normally wait for a reference call to surface shows up during screening instead. A one-way video interview asks every candidate the same structured questions and records their answer in one sitting, no retakes to polish it into something else. You see how someone actually communicates and thinks on their feet, not how a former manager remembers them communicating months later. A talent assessment, whether it’s a Big Five personality profile, a situational judgment test built around how your team actually handles real scenarios, or an environment fit check, adds a second, independent read on temperament and work style that a curated phone call was never designed to produce.

None of this tells you what to decide. Truffle’s AI scores every response against the criteria you define for the role and shows you why a candidate landed where they did. You’re still the one who watches the Candidate Shorts, reads the summary, and decides who’s worth an offer. The difference is that the evidence in front of you was gathered from the candidate directly, under conditions they didn’t get to arrange in advance.

What this looks like without a recruiter

Say you’re the GM of a single-location gym, hiring your third front-desk coordinator this year. The old version of this process: read sixty resumes at night, do three phone screens, call two references each candidate picked for you, and go with your gut. Most of that week goes to confirming things you’d already guessed.

The alternative isn’t skipping verification. It’s reordering it. Resume screening narrows sixty candidates to a shortlist based on the basics. A one-way interview shows you how the top candidates actually communicate and handle a scenario question about a frustrated member, recorded before anyone coached them on what to say. An assessment adds a read on how they handle ambiguity under real conditions. By the time you’re deciding between your final two, you have more direct evidence about how they think and communicate than a handful of reference calls would ever produce.

If you still want to call a reference at that point, the call changes shape. Instead of “tell me about their communication style,” you’re asking a former manager to confirm something specific you already watched the candidate demonstrate. You’re verifying, not discovering. That’s a short confirmation call, not a fishing expedition, and it’s a better use of the phone call you were probably going to make anyway.

Move verification upstream, not into a phone call

Reference checks aren’t going away this year or next. Compliance teams will keep requiring them, and confirming dates and eligibility still matters. But the conversation about whether they’re “worth it” has been asking the wrong question. The real question was never whether to call. It’s where in your process you’re willing to let the candidate choose what you get to see.

Resumes, cover letters, and now references all share the same weakness: the candidate controls the input before you ever see it. As AI makes it easier to polish every one of those inputs into something indistinguishable from the real thing, the only signals that hold up are the ones gathered in the moment, on the record, before anyone had a chance to curate them. That’s not a reason to trust technology more. It’s a reason to trust curated inputs less, and to move your verification to the earliest point where the candidate is still just responding, not performing for an audience they picked themselves.

Frequently asked questions about reference checks in 2026

Should you stop doing reference checks completely?

Not entirely. Keep them for confirming facts: dates, titles, and rehire eligibility. Drop the expectation that a reference call will reveal something new about how a candidate performs or communicates. That evidence should already be in front of you from earlier in your screening.

What should you check instead of calling references?

Look for signals the candidate can’t hand-pick. A one-way video interview shows how someone communicates and thinks through a question in real time. A talent assessment shows temperament and judgment under conditions you control, not conditions a friendly former manager describes secondhand.

Is it legal to skip reference checks?

Yes. Reference checks aren’t a legal requirement in most cases, though some regulated industries and roles have their own verification rules worth checking with your own counsel. Background checks, which confirm identity, criminal history, and license status, are a separate process from reference checks and often carry their own requirements.

How do you add a harder-to-fake signal without slowing down hiring?

Build it into the screening step you’re already running, not as an extra stage after it. A structured one-way interview or a short assessment can run alongside resume review, so by the time you’d normally start dialing references, you already have direct evidence to decide from.