Identity Document Verification Component

Examines a government issued identity document to determine whether it is authentic and unaltered, and whether it belongs to the person presenting it.

Find a section

Home / Screening Components / Identity Document Verification

Overview

Identity document verification examines a physical identity document, usually a driver’s license, state identification card, or passport, and assesses two things: whether the document itself is genuine and unaltered, and whether the person presenting it is the person it describes.

It answers a narrower question than its name suggests. Most screening components confirm information against records held elsewhere. This one starts with the object in the applicant’s hand and works outward from it.

The component became common alongside remote hiring. When a candidate never sits across a desk from anyone, the traditional method of confirming identity, which was a person looking at a card, has no natural moment to happen. Document verification recreates that moment through a photograph and software.

How It Works

The process is short and mostly automated.

CaptureThe applicant photographs the front and back of the document, usually through a phone camera in a guided web session.
ClassificationSoftware identifies which document it is looking at, since a Texas license, an Ohio license, and a passport each have their own layout and features.
AuthenticationThe image is tested against the known characteristics of that document type.
ExtractionName, date of birth, document number, and expiration date are read from the document and returned as structured data.
Face comparisonWhere the service includes it, a selfie is compared against the photograph on the document.

Results usually return in seconds. The output is a set of findings rather than a narrative report, which is part of why it behaves differently from the other components on a screening order.

What the Check Examines

Document authentication looks for the signals that distinguish a genuine document from a forged or altered one.

  • Security features. Modern identity documents carry holograms, microprinting, optically variable ink, and patterns that behave in predictable ways under a camera. Absence of an expected feature is a stronger signal than presence of an unexpected one.
  • Layout and typography. Issuing authorities print to a specification. Fonts, spacing, and field positions that deviate from the known template for that document type and issue year indicate a problem.
  • Internal data consistency. U.S. driver’s licenses carry a barcode on the reverse that encodes the same information printed on the front. Comparing the two catches a large share of alterations, because changing the printed face of a card does not change what the barcode says. Passports carry a machine readable zone with check digits that work the same way.
  • Signs of tampering. Evidence that a photograph has been replaced, that text has been digitally edited, or that the image is a photograph of a screen rather than of a document.
  • Expiration. Whether the document was valid on the date it was presented.

Some services additionally query the issuing authority or an authoritative database to confirm that a document number was issued and remains valid. That step is a different operation from examining the image, and availability varies by document type and jurisdiction.

Face Matching and Liveness

Authenticating a document establishes that the document is real. It says nothing about who is holding it. Face comparison closes that gap by matching a selfie against the photograph printed on the document.

A selfie on its own is easy to defeat, which is why liveness detection accompanies it. Liveness checks distinguish a live person from a photograph, a video replay, a mask, or a synthetic image. Some implementations ask the person to turn their head or follow a prompt. Others analyze a single capture for the characteristics that separate a real face from a reproduction.

Face comparison introduces considerations the rest of the component does not carry. Accuracy varies across demographic groups, a pattern documented repeatedly in federal testing of face recognition algorithms, and the size of that variation differs substantially between vendors. A false non-match falls on the applicant, who is asked to try again or to verify another way.

Collecting a face image also means collecting a biometric identifier, and a number of states regulate the collection, use, storage, and destruction of biometric data, several of them with notice and written consent requirements that attach before collection rather than after. These laws generally apply on their own terms, independent of the federal consumer reporting rules that govern the rest of a background check.

What the Results Look Like

Output is closer to a status than to a record.

VerifiedThe document passed authentication and, where included, the face comparison met the confidence threshold.
Unable to verifyThe check could not reach a conclusion. Image quality is the most frequent cause.
Failed or flaggedA specific problem was identified, such as an expired document, a data mismatch between the barcode and the printed face, or a failed liveness check.

A failure is not the same as a finding of fraud. The distinction matters because these results are frequently automated, and a threshold that has been set to catch more forgeries also rejects more legitimate applicants.

Where It Struggles

The common failure points are mundane and they cluster among people least equipped to work around them.

  • Image quality. Glare, low light, motion blur, and older phone cameras account for most inconclusive results.
  • Physical wear. A document that has been through a wash cycle or carried in a wallet for years may no longer present its security features cleanly.
  • Unfamiliar document types. Coverage is strongest for common U.S. documents. Tribal identification, territorial documents, recently redesigned licenses, and foreign documents are recognized unevenly.
  • Name changes. A document that has not been updated after a marriage or legal name change is genuine, and will still mismatch the name on an application.
  • No document. Not everyone holds current government issued photo identification, and the people who do not are disproportionately low income, elderly, or living without stable housing.

An identity check that cannot complete does not indicate that anything is wrong. It indicates that the check did not work.

How It Relates to Form I-9

These two are adjacent and are often conflated, so the boundary is worth stating plainly.

Form I-9 is a federal legal obligation. Every employer completes one for every employee, it has a required sequence and required timing, and it documents an examination of documents from specific lists. Our Form I-9 page covers it.

Identity document verification is a technology. Nothing requires an employer to use it, and using it satisfies no legal obligation by itself.

The two intersect at one point. The I-9 requires that someone examine the employee’s documents, and a rule permits certain employers to conduct that examination remotely over live video rather than in person. Document verification software is a tool an employer might use within that process, and some I-9 platforms build it in.

Running a document verification is not completing an I-9. The remote examination alternative carries its own conditions, including E-Verify enrollment, retention of document copies, and consistent application across new hires. An identity check that returns “verified” leaves every one of those obligations untouched.

How It Differs From SSN Trace and CBSV

Three components on this site address identity, and each approaches it from a different direction.

Identity Document VerificationExamines a physical document and the face of the person holding it.
SSN TraceSearches commercial data for the names and addresses associated with a Social Security number.
Consent Based SSN VerificationConfirms that a name, date of birth, and SSN match Social Security Administration records.

They fail in different ways, which is the useful part. A stolen but genuine document passes document verification and reveals nothing unusual to an SSN trace. A fabricated Social Security number fails CBSV while the driver’s license in the same file authenticates cleanly. Our pages on SSN Trace and Consent Based SSN Verification (CBSV) cover each in full.

What It Does Not Do

  • It does not establish work authorization. A valid driver’s license says nothing about whether its holder may work in the United States.
  • It does not search any records about the person. No criminal history, no credit, no employment history.
  • It does not prove the document belongs to the presenter. Without face comparison, it establishes only that the document is authentic.
  • It does not detect a document reported lost or stolen unless the service includes an issuing authority query, which many do not.
  • It does not replace the I-9 document examination or satisfy any part of it on its own.

Where It Sits Under the FCRA

Whether the FCRA applies depends on who performs the identity verification.

If a consumer reporting agency (CRA) performs the verification and sends the results to an employer, the verification is generally part of a consumer report. That means the employer must follow the FCRA, including the disclosure and authorization requirements before the check, and the adverse action process if the results affect a hiring decision. Our FCRA Overview covers those requirements in full.

If the employer performs the verification itself using its own software, the FCRA generally does not apply because no consumer report is being created.

Some companies describe themselves as technology providers instead of consumer reporting agencies. That label does not determine whether the FCRA applies. What matters is whether the company collects information about an applicant and provides it to an employer to help make an employment decision.

State biometric privacy laws are different. If a face image is captured, those laws may apply whether a CRA is involved or not.

Applicants usually cannot tell which arrangement an employer is using because the experience looks the same. The difference determines whether the FCRA applies and what legal protections are available if the verification causes an employer to reject an applicant.

Scroll to Top