How Employers Verify Education Credentials: The 2026 Playbook

For most of the last century, the social contract worked. Candidates wrote what they earned, employers trusted it, and the rare liar was an embarrassing anecdote rather than a category. That equilibrium collapsed in three stages: rising wage premiums on credentials, a global market for fabricated degrees, and the arrival of generative AI that produces a competent-looking diploma in five minutes.

The result is a verification industry that did not need to exist in 1995 and is now a default line item in every serious hiring budget. Roughly half of US employers verify the highest degree claimed, with rates substantially higher in healthcare, finance, government, education, and any role that touches a professional license. Our 2026 fraud statistics show the volume that drives it.

Modern education verification is not one step. It is three parallel rails plus a fourth for foreign credentials. Each rail catches a different failure mode, and the slowest rail decides the timeline.

The candidate uploads a PDF or photo of their diploma or transcript. A forensic engine analyses typography, the vector geometry of the institutional seal, registrar signature profiles, PDF producer metadata, and template patterns. It returns a confidence-scored verdict in under a minute.

This rail catches forgeries before any human picks up the phone. It is the only rail that catches AI-generated diplomas, because the registry rails were never designed to ask “is this document real?” They were designed to ask “did this person graduate?” The two questions look the same until the document is fake and the person also graduated, just not with the credential they claim. See our 7 forensic signals guide for the underlying detection logic.

For US institutions, the National Student Clearinghouse operates a service called DegreeVerify, which covers roughly 96 percent of US postsecondary degrees through participating institutions. A query returns the institution, degree, major, and graduation date in seconds. Pricing runs about USD 10 to USD 18 per query at retail, less in volume.

NSC is decisive when it returns a clean match. It is also decisive when the institution is in scope but the candidate is not found, which means the claim is almost certainly false. The ambiguous case is the four percent of schools NSC does not cover, where a missing record means nothing.

For schools outside NSC, the next rail is the institution itself. Most universities now publish a verification portal with a fixed turnaround commitment, usually 3 to 7 business days. A small number of community colleges and older institutions still verify by phone or fax, which extends the timeline by a week.

Background-check vendors maintain a registrar contact database so verification analysts do not have to find the right department under fresh time pressure every time. This is one of the quiet reasons buyers stay with the same vendor for years.

For foreign credentials, the standard step is a credential evaluation by a NACES member: World Education Services (WES), Educational Credential Evaluators (ECE), or one of about a dozen others. The evaluator confirms the institution is recognized in its home country, that the degree is at the level claimed, and how it maps to US academic units.

Costs run USD 90 to USD 220 per report, with turnaround 4 to 20 business days. WES and ECE are bottlenecks during admissions season. For pre-screening, most teams now run a forensic AI check first to discard outright fakes, then commission the evaluation only for the candidates who are otherwise viable. Our foreign transcript verification guide walks the international workflow in detail.

The output of a verification is not a green check mark. It is a structured record: institution name, official institution ID, degree level, major or program, conferred date, candidate name as it appears on the official record, and the source rail that produced each field. Borderline findings carry a disposition note: discrepancy in name spelling, attendance without graduation, attendance dates that do not match the candidate’s claim, or program named differently in official records.

A discrepancy is not automatically a forgery. The most common source of discrepancies in 2026 is candidate self-reporting error: rounded graduation year, conferred degree under a slightly different name, or program merged or renamed by the institution. The verification record exists so a recruiter can resolve the discrepancy with the candidate before it becomes an adverse action.

US employers running an education verification are bound by the Fair Credit Reporting Act when the work is commissioned from a third-party consumer reporting agency, which almost every formal background check is. FCRA requires written consent before the check, accurate dispute resolution after, and a two-step adverse action sequence: a pre-adverse notice with a copy of the report, a waiting period, and the final adverse action notice.

FERPA binds the educational institution rather than the employer. It is the reason a registrar will not release any record without authorization from the student. The university does not work for the employer, and the institution’s legal exposure for unauthorized release is higher than its exposure for refusing to respond.

For roles regulated by professional licensing (healthcare, legal, financial advice, education), an additional rail applies: the licensing board itself maintains a primary-source verification record, which the employer can query directly. A credential that exists in the registrar record but not in the licensing record is a red flag larger than most forgeries.

The 2026 cost stack is roughly as follows, per candidate, in US dollars:

The unit economics matter because the cost of a bad hire in the US sits around USD 240,000 for an individual contributor and substantially more for a regulated role. Education verification is the cheapest credential-fraud control in the hiring stack.

Four failure modes recur. Recognize them, since they describe the cases where a clean verification result is still wrong.

• ·The school exists, the program does not. Diploma mills register a real legal entity at a real address, then issue degrees from invented programs. The NSC knows the entity is real but does not maintain a program catalogue against it.
• ·Identity drift on common names. John Smith graduated. A different John Smith claims the degree. The registrar confirms a John Smith graduated and does not cross-check date of birth unless the employer asks.
• ·Attended but did not graduate.The candidate attended for two years, then transferred or dropped out, but lists a completed degree. The NSC will respond “no record of degree” rather than the more decisive “attendance without graduation” unless asked.
• ·Foreign credential with no equivalency.A diploma is real in its home country but does not map to a US-recognized degree. The candidate listed “equivalent to a master’s” on the resume without an evaluation backing it.

Each of these is caught by a different rail. None is caught by a single rail alone, which is why the playbook is layered.

Two shifts. First, AI-generated diplomas pushed forensic document verification from a final check to a screening step. In 2018, you ran a forensic check on the one candidate who made it to a final-round offer. In 2026, you run it at resume intake on every candidate who reaches a recruiter screen, because the unit cost is now low enough and the false-fake rate high enough to make it the right place.

Second, the AI runs both sides. Some candidates use AI to produce convincing documents. Others use AI to produce convincing claims about training that never happened. The two-rail check (document forensics plus registry confirmation) catches both classes, because they fail on different surfaces. Our AI document verification primer describes the engine that does the lifting.