What the Free Structural Check Actually Covers
The free check runs all five field tests on every reference in your list. It tests for an author field in the expected position for the detected citation style. It tests for a four-digit year in a plausible publication range. It tests for a title fragment of sufficient length to be a real title. It tests for a publisher or journal name. And it tests for an identifier — either a DOI matching the 10.XXXX/ format or a URL beginning with a valid scheme.
The result for each entry is a pass (all five fields present), a warning (one non-identifier field missing or ambiguous), or a flag (identifier absent or malformed). There is no limit on reference list length. A bibliography of 200 entries receives the same complete test as one with 5 entries.
What the Free Check Does Not Cover
The free structural check cannot confirm that a reference resolves to a real document. A reference with a correctly formatted DOI, a complete author, a plausible title, and a named journal passes every structural test — but the DOI may not resolve to anything when entered into doi.org. Confirming resolution requires an HTTP request to the DOI registry, which the in-browser check does not make.
It also cannot confirm that a journal name appears in a recognized database of academic publications, that a reported page range matches a real article, or that an author name corresponds to a real researcher. These deeper verification steps require database access — CrossRef for DOI resolution, PubMed for biomedical sources, or Google Scholar for general academic literature.
Free Structural Check vs. Paid Database Verification — When Each Applies
For most use cases — a student checking an essay before submission, a researcher reviewing an AI-drafted literature review, an editor checking a manuscript's reference list — the free structural check is the right first step. It identifies every reference missing an identifier (the primary fabrication signal) in under five seconds and for zero cost. The references that pass are structurally complete; the ones that fail need manual attention.
Paid database verification tools (like those offered by Semantic Scholar, RefWorks, or institutional library systems) check whether a specific title-author combination exists in an indexed database. This is a deeper test that catches the structurally complete but nonexistent reference — but it requires a database subscription, typically costs per-lookup or per-user, and is slower. The correct workflow for high-stakes submissions is free structural check first, then manual database verification for the flagged subset.