Most teams find out about fake job applications late. A hiring manager spends a second-round interview with someone who can’t explain a single project on their resume, or the person on the final call looks a little different from the one on the first. The application got that far because it read like the good ones, and on a busy role there was no time to read it twice.

After one bad experience it’s tempting to swing the other way, treating every unexplained gap as a lie and every quick application as a bot, until real candidates are handled like suspects.

I’d stop judging fake job applications by how the resume reads. A resume is the one part of an application a scammer gets unlimited drafts of. Everything that has to agree with the resume is much harder to fake: the details around it, and the person who has to talk about the work on a call. So I’d spot fakes by checking the application against itself and then against the person, and stop them with small checks wherever an applicant has to commit to something you can verify.

Know which kind of fake application is in front of you.

Fake applications come in a few kinds, and each gives itself away at a different point in the process, which is where its check belongs.

Plenty of real candidates apply a level above where they are, or describe their last job generously, and that honest stretch is for the first call to judge. Fraud is misrepresenting on purpose, and it comes in four common shapes:

Two kinds cost the most: invented histories and stand-ins, because they can pass a recruiter’s read and reach a hiring manager. Bulk applications mostly cost reading time. The last kind is a security problem as much as a recruiting one, so agree now on who in your security team takes those cases.

Spot fake job applications by reading the details against each other.

The first read is the cheapest place to catch a fake, because an application has to agree with itself, and an invented one has more places to slip. So on the first read, look for details that should line up and don’t, and run the check beside each one:

Each sign in the table has an ordinary explanation, and the check settles which one it is. Write down each check and what it found in one line, for example “profile under a shortened first name, same employers and dates”. A later interviewer then reads an answered question instead of an unexplained doubt.

Put a small check wherever an applicant commits.

Spotting a fake still costs a recruiter’s time, so the cheaper move is to put a check at each point where an applicant has to commit to something you can verify. I’d only add checks real candidates pass easily:

• On the application form, ask one question about work the person has done in a role like this one, written for the role, such as how they handled the last month-end close that went wrong. A real candidate answers it in a few sentences. A tool sending one application to hundreds of roles has no real answer to give, and an answer that echoes the job post is easy to spot on the first read.
• In the job post and the confirmation email, say which checks happen and when, such as video on for interviews, so no check comes as a surprise to a real candidate.
• On one-click application forms, add that same question, so the quickest route in still asks for something written for the role.

Test the claims on the resume during the first call.

Before the first call, pick two claims to test from the resume, for example a project the person says they led, or a tool they list as part of their everyday work.

Someone who did the work can name the people they did it with, and corrects you when you get a detail wrong. Listen for an answer that only repeats the resume in new words, and ask again. I’d follow up on each claim twice, because a first answer can be rehearsed and a second one is much harder to prepare.

Ask for video on and put that in the invitation, with another route for anyone who can’t use a camera, such as a phone call now and video at the next stage. If the picture looks wrong, ask the person to turn side-on or pass a hand across their face. It’s an awkward request and a pass proves nothing, but a face that breaks up when it moves is reason enough for a second call with a colleague.

Write down the two claims and the candidate’s answers in their own words, so a later interviewer can come back to the same claims.

A longer list of signs that a candidate may be fake, across every stage of hiring, sits in this guide to deepfake interviews and fake candidates.

Keep one person in the process from application to offer.

Identity fraud can show up as a handover partway through. One person applies and takes the first call, and a better-prepared person turns up for the technical round, or the reverse.

So give every later round a way to check it’s talking to the same person. The interviewer in the second round reads the first call’s notes and asks about one of the claims tested there. If the account of that project changes, a changed account is evidence the team can act on, which a vague sense that the face looked different never is.

Identity and work-authorization checks belong at the point your employment-verification process and legal team set for them. When they run, ask whoever runs them to confirm that the person they cover is the person your interviewers met.

Decide in advance what a failed check means.

Put the outcome of each failed check in writing before the first case arrives, and share it with everyone who reads applications:

• One check fails and everything else lines up: the recruiter asks the candidate for the specific thing that would clear it, with a reasonable deadline.
• Two checks fail: the recruiter rejects the application and records which checks failed as the reason.
• Anything that looks like an attempt to get inside company accounts or records: stop the process and hand it to your security team before replying.
• A flagged candidate who clears every check: the record says each check was cleared, so no later interviewer inherits the doubt.

With those outcomes written down, a recruiter who feels something is off runs the next check instead of rejecting on a hunch, and they record which check failed where the next reviewer will see it.

Where Metaview fits in stopping fake job applications.

On inbound applications, Metaview’s Application Review assesses each one as the candidate is added, looking for signs of identity deception and application automation, by default and with nothing to set up. Every flag shows how high the risk is and explains it in plain language.

Metaview’s “Shipped at Metaview” episode on the feature shows where the fraud indicator sits in the Application Review table:

A flag is short of a finding of fraud, and Application Review never rejects anyone on its own. In this method, a flag is one sign: a recruiter reads its explanation and runs the checks from the table that fit it.

On a role with hundreds of inbound applications, the flagged ones are where I’d start each day, and the rest of the pile gets its ordinary read.

Checking for a handover between rounds has a weak spot: the final-round interviewer has usually never met the person from the first call. The Metaview Notetaker gives them that earlier call to go back to. Every round it joins runs on the same footing, which Metaview puts to candidates in these words: “If both you and the interviewer consent, you will see a ‘Metaview Notetaker’ appear as a participant during the interview.” Getting that consent is the hiring team’s job, and once it is given, the Notetaker records the call, transcribes it, and builds structured notes from it.

Metaview has used the illustration below to show what those notes look like, with each answer listed under its question:

1

2

1. 1The notes card shows the call’s two participants, one thumbnail carrying a play button.
2. 2The “AI Notes” card opens on “Hi Kevin!” and a line announcing the conversation summary for “Rebecca A.”

Once the Notetaker sits in on each round, Metaview captures every spoken word across all of them, so the final-round interviewer can reopen the first call’s answer about a tested claim and ask about it again.

The headings in those notes come from a template your team picks for each type of call, and an older version of Metaview’s page about the Notetaker showed several, from “Screening Call” to “Final round”:

1

2

1. 1“Screening Call” is the template selected, with its detail and format settings beside it.
2. 2“Topic Highlights” and “Final round” sit behind it, templates for other kinds of call.

Before an offer, several conversations and documents can be combined into one notes document, so whoever signs off reads the attached rounds of one candidate in one place.

Read the past month of check records before adding a check.

The one-line records from the first read are where this method shows whether it is working. Pull every record from last month for an application that was flagged or failed a check, and read them together.

• A flagged application that reached a call with no line beside it skipped its check. Make the line required before anyone books the call.
• A rejection that names only one failed check went past the outcomes you wrote down. Take that case back to the recruiter who made it, with the outcome list open.
• A second-round note that never returns to a claim from the first call leaves the handover unwatched. Add one tested claim to that round’s notes template.
• The same sign clearing again and again, such as profiles under a shortened first name, is an ordinary reading the team already knows. Add it to that row’s “What clears it” column, so the next recruiter clears it in one line.

A month where every flag carries its answer is a record the next interviewer can trust, and only then would I add another check.