A registrar in Hyderabad drops her published case report into a reference manager. The author list comes back as one initial and half of her department's name. The PDF on the journal site is perfect.
The problem is not in the PDF. It is in a separate machine-readable record, deposited alongside the DOI, that she has never seen. That record — not the article — is what reference managers, aggregators and citation-matching services read.
Anyone can open it in about a minute, and once you know which seven fields to check, another minute tells you whether yours is right.
One URL shows you the record your reference manager actually reads
Put any published DOI on the end of this address:
https://api.crossref.org/works/<your DOI>
A live record under our prefix: https://api.crossref.org/works/10.52338/wjsurgy.2020.1001. Open it in a browser and you get the registered metadata served as JSON — the same record software reads, and Crossref metadata is open to everyone.
Your article page and your Crossref deposit are two different objects, and only one is machine-readable
The article page is ours — HTML and a PDF that we host and edit.
The Crossref deposit is a separate XML record, sent when the DOI is registered. It is served from Crossref's registry to anyone who asks — reference managers, discovery services, OpenAlex. The DOI sends a reader to the page; the metadata reaches software without touching it.
Fixing a typo on the article page does not change the deposit. If that split is new to you, start with what a DOI actually is.
Run the 60-second audit: paste your DOI into the Crossref API and read the JSON without being a programmer
- Copy your DOI, starting at
10.— the bare form is the tidiest to paste, though the fullhttps://doi.org/link also resolves — and paste it afterhttps://api.crossref.org/works/in your browser's address bar, with no space or trailing full stop. - Current versions of Firefox, Chrome and Edge render a JSON response as a collapsible, searchable tree; if yours shows one unbroken line, use Ctrl+F to find a field name.
- Everything you need sits inside the
messageobject; the paths below start there.
Check these seven fields in order: title, author names, ORCID, licence, publication date, ISSN and references
| Field | Where to find it | What the Crossref standard expects | Common failure you can spot yourself | Correction notice? |
|---|---|---|---|---|
| Title | message.title[0] | The title as published, one string, without the journal name; Crossref permits a limited set of face-markup tags — <i>, <b>, <em>, <sub>, <sup> — so italics on a species name are correct rather than an error | Truncated at a colon; capitalisation that does not match the article; a tag Crossref does not permit, such as <span> or <p>, left in the string | No — unless you are changing the published title itself |
| Author names | message.author[], each with given and family | One entry per person, personal names split into given and family; organisations recorded as organisations, not people | An affiliation string in the name field as though a department were an author; given and family swapped or merged; a name split mid-word; every entry marked first | No for a spelling or field-split fix; yes if an author is added, removed or reordered |
| ORCID | message.author[].ORCID | A full https://orcid.org/ URI on each author who has one, with authenticated-orcid true only where the iD was confirmed by the author signing in to ORCID rather than typed into a form | Absent entirely; present but carrying the iD of a different researcher with a similar name; an unauthenticated iD nobody has confirmed | No — metadata only |
| Licence URI | message.license[].URL | A licence URI with a content version and a start date — for CC BY 4.0, https://creativecommons.org/licenses/by/4.0/ | No license element at all, so the field is absent; or a URI for a different licence from the one on the article page | No, to add or correct the URI; to change which licence applies, see Copyright & Licensing |
| Publication date | message.published.date-parts | The date of publication, year first, to the precision you have | The deposit date used instead — message.deposited and message.created are Crossref's timestamps, not your publication date; or a bare year where a full date exists | No — metadata only |
| ISSN | message.ISSN and message.issn-type | The journal's ISSN, typed electronic or print, with container-title matching its registered title | A print ISSN typed as electronic; an ISSN belonging to another journal; the journal name abbreviated differently from its registered form | No — metadata only, and it applies journal-wide |
| Reference list | message.reference and message.references-count | One entry per cited work, carrying the cited DOI where one exists, which is what lets Crossref link your citations outward | references-count reads 0 and there is no reference array, so no links run from your article to the work it cites | No — metadata only |
A blank field is not the same as a broken DOI — how to tell an omission from an error
An absent element means it was never deposited. It does not mean the DOI failed: the DOI works if https://doi.org/ plus your DOI lands on the article and the API returns a record.
Separate the two before you write. An omission is a missing field: nothing was sent, something can be added. An error is a field sent with the wrong content in it — a swapped surname, another journal's ISSN. Saying which you found saves a round of emails.
What we can and cannot tell you about our own deposits under prefix 10.52338
We are a Crossref member, and every published article carries a registered DOI under prefix 10.52338. Content is produced in JATS XML, and the metadata is exposed by OAI-PMH at https://directivepublications.org/oai, which answers ?verb=Identify.
What we will not do is claim in advance what is in your particular record. Pull a spread of live 10.52338 records, as we did before writing this, and the gaps are visible: many carry no license element; ORCID iDs appear on very few; reference lists are not deposited, so references-count reads 0; and on older records an affiliation string has often landed in an author name field, the failure in the opening paragraph.
We will not quote coverage figures or promise a date by which that changes. The template below exists because these gaps are real, and because the record is open for you to check at any time.
A clean record feeds OpenAlex on its timetable, not ours — Google Scholar reads your article page instead, and neither route will ever put you in PubMed, Scopus or Web of Science
OpenAlex builds on Crossref metadata, so a corrected deposit is the right first move. When OpenAlex next refreshes is theirs to decide, not something a publisher can date.
Google Scholar works differently. Scholar builds its records largely by crawling the article page and reading the citation meta tags in its HTML, not by reading Crossref. A Crossref fix is therefore not a reliable route to changing what Scholar shows; the page is where to look. See choosing titles and keywords for discoverability.
One boundary: no amount of metadata quality places an article in PubMed, PubMed Central, MEDLINE, Scopus or Web of Science. Those services select journals through their own evaluation; Directive journals are in none of them, and a perfect record does not change that.
Metadata-only fixes need no correction notice; changing the article itself does
Correcting a deposited field does not alter the published article, so it needs no published notice; once the deposit is updated, the corrected record travels out from Crossref.
The moment a change touches what the article says or who it credits — a revised conclusion, an author added or removed — it stops being metadata and becomes a correction to the version of record, handled under Corrections, Retractions & Withdrawals.
Send this metadata correction request, with the four things that let us act on it without a chase
Subject: Metadata correction request — DOI 10.52338/[...]
Dear Editorial Office,
I am the corresponding author of "[article title]", DOI [10.52338/xxxxxx.YYYY.NNNN].
The Crossref record for this DOI contains an error in the [author names / title / licence / publication date / ISSN] element.
It currently reads: [paste the exact string as it appears at https://api.crossref.org/works/ followed by the DOI]
It should read: [the exact corrected string, verbatim, with nothing else changed]
This is a metadata correction only. The content of the published article is unchanged, no author is being added, removed or reordered, and the conclusions are unaffected. I am writing on behalf of all authors.
Please confirm when the deposit has been updated.
[Name], corresponding author
If your request changes the article's meaning rather than the record describing it, send it under Corrections, Retractions & Withdrawals instead.
Checking the record is part of what the corresponding author is on the hook for; our licensing terms are at Copyright & Licensing. See also what a DOI is and article versions explained.