45% Coach Their References. Brief Yours on the Tailored Resume Instead

At the reference stage, your former manager probably won't get a phone call. They'll get a structured survey with the same questions every other finalist's referees answer, and the platform checks their answers against those referees and against a list of fraud signals.

Dice's piece on the resume-screening shifts that will raise the bar in 2027 says AI screeners now verify skills, check references, and cross-reference your GitHub. The fix isn't scripting your references. It's briefing them on the exact resume you sent and letting them choose their own words.

What is an automated reference check?

An automated reference check is software that sends each of your referees the same online survey, runs fraud checks in the background, and gives the recruiter a report that compares your referees with other candidates'. You do the legwork of collecting and chasing referees. The software handles the survey and the scoring.

Harver says its tool works "by placing the responsibility of chasing down references onto the candidate." You type in names, emails, and phone numbers. Checkster sends a 2-minute mobile survey, and Harver says the check flags up to 5% of candidates and returns references within 24 hours.

The platforms you're most likely to run into:

The differences between them matter less than what they share. Every one of them turns your referee into a data point.

What do references get asked?

In an automated check, your referees answer a standardized questionnaire, and every candidate's referees for that role get the same questions. The point is comparison: your referee's answers get read next to everyone else's.

Harver says its question templates are validated by independent I/O psychologists, so a recruiter isn't making them up on the fly. Xref describes its report as showing "how the candidate performed against other individuals referenced."

Your referee is graded on a curve. When every finalist's referees answer the same prompts, a reference who writes "great to work with, very reliable" sits right next to one who writes "rebuilt our billing reconciliation and cut month-end close by a week." The first answer isn't negative. It just loses.

Structured surveys strip out the tone and rapport a phone call used to carry. What's left is content, and specific content wins. So the job isn't finding a referee who likes you. It's finding one who remembers the work, and making sure they remember the right work.

That means picking the person who actually saw the skill the posting ranks highest, even if they aren't your most senior contact. If the role mostly wants data pipeline ownership, the staff engineer who reviewed your pipeline code every week beats the VP who saw you in two quarterly reviews. Working out which requirement carries the most weight is the hard part. Refolk scores how well you fit each posting requirement by requirement, which shows you which claims a referee most needs to back up.

The "41% send fake references" stat is misquoted

GCheck's survey doesn't say 41% of job seekers send fake references. The 41% is people who had a friend or family member pose as a professional reference. Only 27% listed outright fake references, and the biggest number in the survey is coaching.

The full breakdown:

The generational split is wide. GCheck found Gen Z candidates listed fake references at 40%, compared with 7% among Baby Boomers. A separate report on automated reference checks ties the behavior to market pressure (72%) and to an expectation that verification will be weak (53%).

That last figure is the one that matters. More than half the people cutting corners assume nobody checks, and automated reference checks exist to prove them wrong. Since coaching is the most common move, the tools are tuned to catch it, which means a well-meaning candidate who over-prepares a referee can look exactly like someone gaming the system.

How AI reference check surveys spot a coached referee

These tools don't judge whether your referee is telling the truth. They look for signs that you and your referee are too closely linked: same device, same network, same words. Crosschq and HiPeople both document the signals they check.

What gets flagged:

• Shared IP address. Xref has described a customer case where "the reference had been completed from the same IP address as the candidate."
• Device fingerprint matches. HiPeople fingerprints the device and IP of both the candidate and each reference.
• LinkedIn mismatches. Crosschq flags LinkedIn work histories that don't match.
• Disposable email or VoIP numbers. Crosschq flags both. HiPeople also checks how old a phone number is and who the provider is.
• Answers that echo the resume word for word. This is a known sign of a stand-in or coached referee. If your manager writes your bullet back to the recruiter verbatim, it reads as a script, not a memory.
• Family posing as managers. Harver says Checkster can catch candidates passing off family members as former managers.

Crosschq and HiPeople both note that a shared office network can produce an innocent match. That helps if you and your referee still work at the same company, but you don't want to be explaining it.

Reference check after layoff: when a real referee looks fake

A legitimate referee who was just laid off can trip the same signals as a fake one. A former manager answering from a personal Gmail on a new mobile number, with a LinkedIn profile that still lists the old job, matches three fraud signals at once.

The October and November cuts made this common. Netflix announced cuts of about 6% of staff, close to 1,000 people, in October. Canva cut 22% of its staff (1,271 people), Fiserv cut about 3,000 employees and contractors under its new CEO, and Block made November cuts without disclosing a headcount. The disclosed cuts at Canva, Netflix, and Fiserv add up to roughly 5,271 people. Many of them managed someone who's now job hunting and needs them as a reference.

The pool of likely referees is big. Every manager who changes jobs or gets cut becomes a referee whose LinkedIn, email, and phone may stop matching what you typed into the reference form.

On work email: some platforms lean on it. Veremark validates work emails, and RefNow verifies referees through LinkedIn and Yoti. There's no evidence that Harver, Xref, or HiPeople strictly require a corporate address. But a personal email combined with a stale profile and a new number is exactly the pattern the fraud checks look for.

Location can also sink a reference with nobody at fault. Checkr's reference checks only work for US background checks, and a reference fails if their phone number is unsupported, such as a non-US number. Many of Canva's laid-off managers are in Sydney. If a Checkr-run check reaches one of them, it can come back as no response.

How to fix it before the survey goes out:

1. Ask the referee to update LinkedIn first. Their work history should show the role they held when they managed you, with accurate dates.
2. Use the email and phone they'll actually check. Ask them which ones, and don't guess. A disposable or brand-new number is a flag.
3. Have them answer from their own device on their own network. Not at your place, not at a coworking space you both use, not on your laptop.
4. Ask the recruiter which platform they use if your referee is outside the US. If it's Checkr, offer a US-based alternative or ask for a manual check.
5. Use official outplacement channels where they exist. Canva says its talent acquisition team will also help laid-off staff with resumes and interview practice. For a referee coming out of Canva, that's a clean path to getting their profile in order before your reference request lands.

How to prepare your references using the tailored resume

To prepare your references, send each one the exact resume version you submitted for this posting. Point to the two or three claims that matter most, remind them of the projects and outcomes behind those claims, and don't write their answers.

This is where most candidates slip. You tailored the resume, and the bullets that got you through the screen were rewritten to match this posting's language. Your referee remembers you from two jobs ago and has never seen this version. If the employer's system lines the referee's survey up against the skills on your resume, you want both describing the same work.

A brief that holds up:

1. Send the tailored copy, not your master resume. If you're applying to a lot of roles, keeping track of which version went where is real work. That's what Refolk does: it writes your resume from your own history and tailors a version for each posting, so pulling up the copy that went to this employer is quick.
2. Name the claims to cover. "The posting leans hardest on vendor migrations and stakeholder management. Those are the two things on my resume they'll ask about."
3. Jog their memory with specifics. "Remember the Q3 payments cutover? I owned the rollback plan." Give them the project, the date, and the outcome. Let them describe it in their own words.
4. Never send wording. No sample answers, no bullet points to paste. Verbatim echoes of your resume are a known fraud signal, and 45% of candidates coaching their references is exactly why the tools look for it.
5. Tell them the format. It's a short mobile survey (about two minutes on Checkster), not a call. Referees who expect a chat tend to write one-liners.
6. Warn them they may be recruited. Checkster asks referees whether they want to hear about jobs. Over 50% of references opt in, creating at least 3 new passive candidates per reference check. A recently laid-off manager is especially likely to say yes. That's fine, but they should hear it from you first.

FAQ

Can I write out what my reference should say?

No. Coaching is the most common form of reference fraud (45% of job seekers in GCheck's survey), and these platforms are built to catch it. Answers that repeat your resume word for word are a known sign of a coached or stand-in referee. Send the tailored resume and remind them of the projects behind each claim, but let them write their own answers.

What if my former manager was laid off and has no work email?

There's no evidence that Harver, Xref, or HiPeople strictly require a corporate email, though Veremark validates work emails and RefNow verifies referees through LinkedIn and Yoti. The real risk is the combination: a personal email, a brand-new phone number, and a LinkedIn profile that still shows the old job can look like a fake referee. Ask them to update LinkedIn, use the email and phone they actually check, and answer from their own device.

Does Checkr's reference check work outside the US?

No. Checkr's reference checks are only available for US background checks, and a reference fails if their phone number isn't supported, such as a non-US number. Checkr also asks for a full first and last name for each reference, allows up to three of each type, and tries each one once per business day for four days. If your best referee is abroad, raise it with the recruiter before the check starts.