# The Identity-Check Pass Standard, Six Criteria Graded per Live Application

*Grade each live application pass, fix, explain, or fail on six identity-check criteria, so two graders reach the same verdict before the next screen is booked.*

- Canonical URL: https://www.refolk.ai/candidates/guides/identity-check-pass-standard
- Pillar: Applying at volume
- Format: Standard
- Published: 2026-10-30
- Last reviewed: 2026-10-30
- Reading time: 19 min
- Keywords: how to pass identity verification for a job interview, flagged as fake candidate job application, VPN during video interview flagged, show ID on camera job interview, name on application doesn't match ID verification, employer identity check before interview what to expect

## Key takeaways

- Greenhouse compares three application fields against the CLEAR identity result (last name, email address, and phone number), and a mismatch in any one of them can fail the verification.
- A VPN is the only common candidate habit Greenhouse documents as an automatic failure, so turning it off before every verification and interview session removes a whole category of outcome.
- Greenhouse reports its Real Talent add-on flags one in four candidates for review and fails one in 33 outright, which, assuming independent checks, means roughly a 94% chance of at least one flag across 10 such applications.
- A nickname, a married versus maiden name, or a different portal email is a mismatch Greenhouse says may be explainable, so the fix for a submitted application you cannot edit is a short note to the recruiter.
- In Refolk's index, 431 US recruiters or talent acquisition staff list Background Checks as a skill and only 3 list Identity Verification, which suggests candidates meet an automated vendor check before a person sees any context.
- A failed identity verification does not automatically disqualify a candidate under Greenhouse policy, and an expired link, an unsupported location, or declining are not counted as failures.

This is the standard I use to make sure an employer's fraud and identity checks see a real person before anyone judges the work. It is for job seekers with dozens of applications out and interviews starting. It gives you six pass criteria, a grading rule two people would apply the same way, a per-application procedure, and a checklist to run before each invite is booked.

## What does an employer identity check actually compare?

At employers that run one, an identity check compares your last name, email address, and phone number on the application with what a verification provider returns, then adds network, device, and camera signals on top. Identity verification here means matching a selfie to your government ID and then comparing the verified result with your application, field by field.

Greenhouse is the only major applicant tracking system (ATS, the software employers use to receive and route applications) whose documentation on this I could find. Its verification runs through CLEAR, and the mechanics are specific:

- **What CLEAR sends back.** Pass or fail, full name, email, phone, and IP address. Your selfie and ID images stay with CLEAR.
- **What gets compared.** Greenhouse cross-checks last name, email address, and phone number against the application. A mismatch in any of these can fail the verification.
- **When it fires.** In some offices or departments it runs as part of the application itself. Otherwise it runs at a later stage the recruiter chooses.

On top of the ID check sit fraud-detection layers that score the application and the interview itself:

- **Application signals.** Greenhouse's Real Talent add-on surfaces structured fraud risk signals from phone, email, IP, and location.
- **Change signals.** Sardine lets customers flag when a candidate's location, device, browser, or operating system changes between application and interview. Its starting rules include New Email, Poor Phone Name Match, and Risky IP Country.
- **Live-interview signals.** Sardine checks for virtual cameras, injected video feeds, and face-swapping tools. An injected feed is a video stream that software pushes into the call in place of a physical camera's output. BrightHire checks for VPN, proxy, Tor, and unexpected geolocation.

#### Where identity signals are collected

1. **Application data** - Name, email, phone, IP, and location scored for risk at submission
2. **Identity verification** - Selfie matched to government ID, then last name, email, and phone compared with the application
3. **Live interview** - Geolocation, VPN, device changes, virtual cameras, and injected feeds checked on the call
4. **Pre-offer round** - An in-person round or on-camera ID check before any offer is made

*Each layer adds signals on top of the one before it, so a mismatch at the application layer follows you into every later round.*

The interview platform changes what gets checked. BrightHire documents different network coverage by platform:

| Platform | Signals BrightHire documents |
|---|---|
| Zoom | Fullest set |
| Microsoft Teams | Geolocation and VPN |
| Google Meet | Deepfake detection only |

I found no public field-level documentation for Ashby, Lever, Workday, or iCIMS. So I grade every application against Greenhouse's documented fields, the strictest baseline that is published. Treat the others as unknown, not as lenient.

## How likely is a legitimate volume searcher to be flagged?

If you apply to many employers using this kind of add-on, at least one review flag is close to certain, so trying to avoid every flag is the wrong goal. Greenhouse reports that Real Talent has run more than 10.2 million fraud checks, flagging one in four candidates for review and failing one in 33 at verification outright.

#### Real Talent fraud checks, per candidate

| Stage | Figure | Note |
| --- | --- | --- |
| Fraud checks run | 10.2 million+ | All Real Talent checks reported by Greenhouse |
| Flagged for review | 1 in 4 | Sent to a human reviewer, not a finding of fraud |
| Failed at verification | 1 in 33 | Outright failure at the identity step |

*Most flagged candidates are not failed; a review flag sends you to a human, while an outright fail is about one in 33.*

At volume, the per-check rate compounds. The table below is my own arithmetic, assuming each check is independent:

| Applications at Real Talent employers | Chance of at least one review flag | Chance of at least one outright fail |
|---|---|---|
| 1 | 25% | 3.0% |
| 10 | 94% | 26% |
| 40 | ~100% | 71% |

Two limits on that table. First, independence overstates the risk, because flags concentrate on the same people, and a clean candidate is less likely to be flagged than the average. Second, only employers that buy the add-on run these checks, and Greenhouse does not publish a false-positive rate (how many flagged candidates turn out to be legitimate). Nobody publishes how often VPNs, nicknames, or device changes show up among legitimate candidates either.

Reviewers also start out suspicious. In Greenhouse's internal machine learning pilot, more than 35% of applicants were flagged high-risk, and 91% of those the security team reviewed were confirmed fraudulent. Pools for remote technical roles contain enough fraud that a reviewer in one tends to read a flag as fraud. The wider base rates explain why:

- A Gartner survey of 3,000 job candidates found 6% admitted to participating in interview fraud.
- Checkr's survey of 3,000 US managers found 40% had interviewed a candidate later revealed to be using a fake identity, up from 31% the year before, and 67% had encountered proxy fraud, up from 35%.
- An FBI public service announcement says North Korean IT workers use AI face-swapping in interviews.
- Coda Search/Staffing reports in-person interview requests from its clients rose from 5% to 30%.

> You cannot avoid the flag at volume. You can make every flag resolve the same way, in your favour.

## The six pass criteria

An application passes this standard when it meets all six criteria below, each checked against evidence and not memory. Each one maps to a signal that a documented check reads, and each has a way it can mislead the grader.

| Criterion | Pass when | When it lies |
|---|---|---|
| 1. Name match to ID | Last name on portal equals ID, character by character | Nickname or maiden name on a real person |
| 2. Location vs connection origin | IP region and device match stated city and the application device | VPN or carrier routing moves you |
| 3. Phone and email consistency | One email, one carrier phone in your legal name, everywhere | Search-only email or VoIP number |
| 4. Profile age and footprint | Email and accounts predate the search | New inbox made for job hunting |
| 5. Camera and background | Native camera, real background, no blur | Effects software reads as injected feed |
| 6. ID-on-camera readiness | Unexpired ID, readable on camera, matches all fields | Glare, expired ID, or a name drift |

### What each criterion proves

1. **Name match** proves the person verifying is the person who applied. Greenhouse compares the last name, so a surname mismatch is the most serious kind.
2. **Location and connection origin** proves you are where you said you are. An IP address resolves only to an internet provider's network or general region, never a street address, so "match" means the region is consistent with your stated city. Device continuity sits here too, because Sardine flags location, device, browser, and OS changes together.
3. **Phone and email consistency** proves the contact details belong to one person. These are two of the three fields Greenhouse compares, and Sardine's Poor Phone Name Match rule tests whether a number fits the name.
4. **Profile age and footprint** proves the identity existed before the search. The documented signal is Sardine's New Email rule. How the age of other profiles is weighed is not publicly established, so the conservative grade is: do not create new identity accounts mid-search, and log any you had to create.
5. **Camera and background** proves the video is captured live by real hardware. Virtual cameras and injected feeds are checked directly, and blur is treated as an artifact signal. A ban on virtual backgrounds is not established in any source I found; the risk is the software that produces them.
6. **ID-on-camera readiness** proves you can complete a live check on request. Guidance summarised from the FBI recommends a live on-camera identity check for every remote candidate at a fixed stage.

Here is how Greenhouse documents the outcome of ordinary habits:

| Habit | Documented outcome |
|---|---|
| Nickname on application, legal name on ID | Mismatch that "may be explainable" |
| Married name vs maiden name | Mismatch that "may be explainable" |
| Different portal email | Mismatch that "may be explainable" |
| VPN during verification | Fail |
| Link expired or unsupported location | Not a failure |
| Declined | Not a failure |

> **Rule:** VPN off, every session, no exceptions
>
> A VPN is the only common habit Greenhouse documents as an automatic CLEAR failure. Turn off privacy and corporate VPNs before every verification and every interview, then confirm with an IP lookup.

## How do two people grade the same application the same way?

Two graders agree when every criterion gets exactly one of four verdicts, each verdict has a written trigger, and every grade cites the evidence it was read from. A grade from memory counts as ungraded.

| Verdict | Trigger |
|---|---|
| Pass | Condition met and evidence recorded |
| Fix | Condition not met, and editable before the next screen |
| Explain | Condition not met, not editable, recruiter note sent and logged |
| Fail | Condition not met, not editable, no note sent |

Grading rules that keep the verdicts stable:

- **Compare characters, not impressions.** "Cameron" and "Cam" is a mismatch, even if it is obviously you.
- **Grade the portal, not your resume file.** The compared fields are what the employer's system holds, including the portal login email.
- **Criterion 2 is graded per session.** A pass at application does not carry to the interview. Re-run the IP lookup each time.
- **One Fail blocks the invite.** An application is ready for the next screen only when all six criteria read Pass or Explain.
- **Unknown ATS is graded as Greenhouse.** If you cannot tell which system an employer uses, apply the Greenhouse fields.

The usual cause of drift across 40 applications is retyping the header by hand. If [Refolk](/candidates) tailors your resume and cover letter to each posting, both come from one record of your history, so the name, email, and phone stay the same across every version. You still have to grade the portal fields yourself.

**Per-application identity grading sheet**

```
Application: <company> / <role> / <ATS if known>
Stage and next session: <stage> / <date>
Device at application: <device> / <OS> / <browser>

C1 Name match         PASS | FIX | EXPLAIN | FAIL   Evidence: portal last name "<...>" vs ID "<...>"
C2 Location + device  PASS | FIX | EXPLAIN | FAIL   Evidence: IP lookup region "<...>" vs stated city "<...>"; device plan "<...>"
C3 Phone + email      PASS | FIX | EXPLAIN | FAIL   Evidence: portal email "<...>", login email "<...>", phone "<...>"
C4 Footprint age      PASS | FIX | EXPLAIN | FAIL   Evidence: email in use since "<...>"; new accounts "<none or list>"
C5 Camera setup       PASS | FIX | EXPLAIN | FAIL   Evidence: native camera, no effects software, real background
C6 ID readiness       PASS | FIX | EXPLAIN | FAIL   Evidence: ID expiry "<...>"; readable on test call yes/no

Recruiter note sent: <date or n/a>
Verification status: <not sent / pending / verified / resend requested>
Ready for next screen: YES only if no FAIL
```

*One block per live application; paste into your tracker and fill every Evidence field before assigning a verdict.*

## The pre-screen procedure, run per application

Run the first step once, steps two to four across your whole pipeline, and steps five to seven per employer as each invite arrives. Each step has a done condition you can check.

#### Identity-check pass procedure

1. **Write down your canonical identity** - Record your legal name exactly as on your ID, one email, and one phone, in about 30 minutes, once. Done when every application uses the same last name, email, and phone, the three fields Greenhouse compares.
2. **Audit every live application** - Check each for a nickname or maiden-name mismatch and a different portal email, about 5 minutes each. Done when a mismatch log exists and every uneditable mismatch has an explanation sent to the recruiter.
3. **Normalise the network** - Turn off any VPN before every verification and interview session and use the same home connection each time. Done when an IP lookup shows a region matching your stated location, since a VPN is an automatic CLEAR fail.
4. **Fix the device plan** - Use the same device type for applying and interviewing, or plan the change, because device and OS changes between stages are flagged. Done when the device for each stage is written into your tracker.
5. **Run each verification link promptly** - Complete the CLEAR or similar check within its window from a supported location, in 2 to 10 minutes. Done when the status shows verified; for an expired link or unsupported location, ask for a resend.
6. **Rehearse on camera** - Test with a physical camera, no virtual camera software, a real background, and an unexpired ID ready to hold up, in about 15 minutes. Done when the ID reads clearly on camera and a head turn or hand pass looks clean.
7. **Prepare for the in-person or pre-offer check** - Keep your identity documents ready for a final round that may be in person. Done when the ID matches every application field for that employer.

Sources differ on when checks happen. Greenhouse can trigger verification inside the application or at a stage the recruiter picks. Other providers describe checks before, during, and after the offer. The common pattern is a remote recruiter screen and remote early rounds, with one in-person round added before an offer. That is why step seven exists even for fully remote roles.

On timing: CLEAR says verification takes seconds if you are already a CLEAR user. If you are not, expect it to take longer; no figure is published, so leave yourself time and do not start it two minutes before an interview.

## How this goes wrong: false positives and failure modes

Most identity-check failures among real candidates come from ordinary search habits, not from anything suspicious, and each one has a specific local check. A false positive here means a legitimate person producing a signal that looks like fraud.

1. **Name match.** A nickname or maiden name produces a mismatch, and the worst case is a real person failing on the last-name field. Check: compare the last name in each portal against your ID, character by character, including hyphens and accents.
2. **Location versus IP.** A corporate VPN, a privacy VPN, or carrier routing can place you in another region. BrightHire itself notes that location obfuscation is common in interview processes. Check: open an IP lookup page right before the session and compare the region with your stated city.
3. **Email consistency.** A search-only inbox that differs from your portal login triggers a mismatch, and Sardine runs a New Email rule. Check: one email for every application and every portal account.
4. **Phone.** An out-of-state or VoIP number can trip a Poor Phone Name Match. Whether the area code alone triggers it is not established. Check: use a carrier number registered to your legal name.
5. **Device change.** Applying from a laptop and interviewing from a phone creates a device and OS change event. Check: log the device used at each stage, and keep it the same where you can.
6. **Camera.** Virtual camera software, such as the kind used to add background effects, can read as an injected feed, and blur is treated as an artifact signal. Check: select your device's native camera in the meeting settings and turn effects off.
7. **Liveness gestures.** Liveness gestures are actions such as waving a hand in front of your face or turning your head, used to expose a synthetic video. Sources say the hand-wave and "three finger" tricks no longer reliably catch a real-time deepfake. So passing one proves little, and an interviewer may escalate to an ID check anyway. Check: be ready for the escalation, not just the gesture.
8. **Reading the base rate wrong.** "Flagged for review" is not a finding of fraud, and Gartner's figure that one in four profiles will be fake is a forecast, not a measurement. Check: treat a flag as a request for context, and supply it.

Remote-desktop tools such as AnyDesk and TeamViewer appear in FBI and Mandiant fraud indicators. If one is installed for legitimate reasons, close it before any interview session.

> **Watch out:** Do not over-fix by masking
>
> The instinct after a flag is to hide more: a fresh inbox, a VPN, a blurred room. Every one of those is itself a documented signal. Fix by making the record consistent, not by making it quieter.

When a mismatch cannot be edited, the documented path is a human follow-up. Greenhouse says a recruiter may send a new verification request to a candidate who fails some checks but appears legitimate. Make that easy:

**Mismatch explanation note to a recruiter**

```
Subject: Identity verification for <role>, <your legal full name>

Hi <recruiter name>,

Before my identity verification for the <role> application, a quick note so the result makes sense:

- My application lists my name as "<name on application>". My government ID reads "<legal name on ID>". <One clause on why: nickname / married name / former name.>
- I applied with <email A>. My portal login is <email B>. Both are mine; <email A> is the one to use.

I completed verification from <city>, on my home connection, with no VPN. If anything came back inconsistent, I am glad to redo it or show my ID on camera at the start of our next call.

Thanks,
<legal full name>
<phone on application>
```

*Send before or right after verification; keep it to facts the recruiter can check against your ID.*

## Who runs the check, a recruiter or a vendor?

The first identity check you meet is usually automated and run by a vendor, not by a recruiter who can read context. Refolk's index of professional profiles points the same way.

**431 to 3 - US recruiters or talent acquisition staff listing Background Checks versus Identity Verification as a skill**

In Refolk's index, Background Checks is listed about 144 times as often as Identity Verification in the US.

| Segment in Refolk's index | Count |
|---|---|
| US recruiters/TA, Identity Verification | 3 |
| UK recruiters/TA, Identity Verification | 0 |
| US recruiters/TA, Background Checks | 431 |

In a 25-person sample from the 431, the top employers were the US Army (5) and Wells Fargo (2). Background checks are an established recruiting skill. Identity verification almost never appears as one.

My reading is that verification sits with outside vendors such as CLEAR, Sardine, and BrightHire, so you meet an automated threshold before anyone sees context. That is an interpretation of a skills field, not a measured finding about who presses the button. The practical effect is the same either way: your application fields have to be consistent with no explanation, because the first reader is a comparison rule.

## What does a failed check mean, and what are your rights?

Under Greenhouse's stated policy, a failed identity verification does not automatically disqualify you from proceeding, and declining verification is not counted as a failure. An expired link and an unsupported location are not failures either. The documented recovery is a resend after a recruiter follow-up.

What I can and cannot tell you about legal limits:

- **Illinois biometrics.** If a tool extracts face geometry or a voiceprint, Illinois' biometric law (BIPA, 740 ILCS 14) imposes its own written-notice, written-consent, and retention-schedule duties.
- **Illinois video interviews.** The Illinois AI Video Interview Act (820 ILCS 42) requires deletion of interview videos within 30 days when the applicant asks.
- **What Sardine says it measures.** It frames its signals as tied to fraud indicators such as device reuse, automation behaviour, impersonation, and location inconsistencies, "not who a candidate is".
- **Not established.** I did not find any regulator guidance, from the EEOC or anyone else, saying that accent or location alone must not be treated as a fraud signal. If you think a flag rests on either, ask the employer which signal triggered it and get advice for your own jurisdiction.

> **Note:** Declining is not failing, but it is not free
>
> Greenhouse does not count a declined verification as a failure. The process consequences of declining are not documented, so if you decline, offer an alternative such as an on-camera ID check in writing.

## The per-application checklist, and keeping it current

Run this checklist before each interview invite is booked. An application is ready when every item is true. For anything you cannot make true, the grading sheet should show Explain, with a note sent.

#### Before booking the next screen

- [ ] The last name in this employer's portal matches my government ID character by character.
- [ ] The email on the application and my portal login email are the same address, or the recruiter has a note explaining both.
- [ ] The phone number on the application is a carrier number registered to my legal name.
- [ ] No new email or identity account was created for this search, or any that was is logged.
- [ ] My VPN is off and an IP lookup run today shows a region consistent with my stated city.
- [ ] The device I will interview on is the one logged for this application, or the change is planned and noted.
- [ ] Any verification link for this employer shows verified, or a resend has been requested for an expired or unsupported-location result.
- [ ] The meeting app is set to my native camera, with background effects and blur off and a real background behind me.
- [ ] My government ID is unexpired, within reach, and readable when held to the camera on a test call.
- [ ] No remote-desktop tool is running on the interview device.
- [ ] The grading sheet for this application has evidence in every field and no criterion graded Fail.

Keeping the standard current takes three habits:

- **Re-grade criterion 2 every session.** Networks, carrier routing, and VPN auto-connect settings change without telling you.
- **Re-check the source documents when outcomes surprise you.** The field list, the VPN rule, and the not-a-failure outcomes come from Greenhouse's support articles on identity verification with CLEAR. If an employer's result does not match this guide, read the current articles before assuming the rules hold.
- **Re-grade when anything in your identity record changes.** A new phone, a name change, or a move resets criteria 1 to 3 across every live application, not just the next one.

When you add an application, grade it as you submit it. Refolk drafts each tailored version from the same canonical record, so the remaining manual work is the portal audit and the per-session network and camera checks.

## Frequently asked questions

### What happens if I fail an identity verification for a job application?

A failed verification does not automatically disqualify you under Greenhouse's stated policy. If you fail some of CLEAR's checks but appear legitimate, the recruiter may follow up and send a new verification request. Expired links, unsupported locations, and declining are not counted as failures at all. Your best move is a short, factual note to the recruiter explaining any mismatch, such as a nickname or a different portal email, and asking for a resend.

### Can I use a VPN during a video interview or identity check?

No. Greenhouse documents that a CLEAR verification attempted over a VPN is marked as a failure, and interview-layer tools such as BrightHire check for VPN, proxy, Tor, and unexpected geolocation. Turn off privacy and corporate VPNs before every verification and interview session, then run an IP lookup to confirm the region it shows matches the city you stated on your applications.

### My application uses my nickname but my ID has my legal name. Will verification fail?

It can produce a mismatch, but Greenhouse lists a nickname on the application versus the legal name on the ID as a mismatch that may be explainable, using Cam versus Cameron as its example. Last name is the compared field, so a married versus maiden surname is the higher risk. Fix the name in any portal you can still edit, and send the recruiter a one-line explanation for the rest.

### Will I have to show my ID on camera in a job interview?

You may. Guidance summarised from the FBI recommends a live on-camera identity check for every remote candidate at a fixed stage, and some employers run a selfie-plus-ID check through CLEAR instead. Keep an unexpired government ID within reach for every remote round, rehearse holding it so the text is readable, and make sure the name on it matches every application field.

### What does the employer actually see from a CLEAR identity check?

According to Greenhouse, CLEAR shares the pass or fail result, your full name, email, phone number, and IP address with Greenhouse. Your selfie and ID images stay with CLEAR. That is why the name, email, and phone you put on each application matter: they are compared against what CLEAR returns, and the IP address shows where you connected from.

### Can I use a virtual background in a job interview?

A blanket ban on virtual backgrounds is not established in any source I found. The documented risk is mechanical: fraud tools check for virtual cameras and injected video feeds, and blur is treated as an artifact signal. Virtual camera software of the kind used to add background effects can look like an injected feed. Use your device's native camera and a real, tidy background.

---

*From the Refolk guide library. I revise these guides rather than replacing them, so the current version is always at https://www.refolk.ai/candidates/guides/identity-check-pass-standard*