Why this signal matters for Google
A list of titles is not proof. "Author of several scientific articles" says nothing: not where, not when, not about what. The reader cannot go and look. Knowing how to cite your publications properly fits in one line and checks out immediately: the exact title, the name of the journal or publisher, the date, and an address that opens the original. It is the same standard you apply to the sources you cite in your articles, turned on your own work this time. The difference between the two versions is not a matter of style. It is a matter of what a stranger can open in a new tab.
Google has humans run that check. Evaluators rate page quality against a handbook Google writes and anyone can download, the Quality Rater Guidelines, which I shorten to QRG below. Section 3.4 asks them to look at the experience and expertise of the person who wrote the content. Section 2.5.2 asks them to work out who is responsible for the site and for what it says. A verifiable publication answers both questions at once. It names an author, and it shows that somebody outside agreed to publish the work: a journal, a publisher, or a peer review panel, meaning other specialists who read the text before it went out. That outside filter is what gives the signal its weight. Nobody self-publishes into a peer-reviewed journal. Readers hold the same standard: in the Knight Foundation and Gallup survey (2018), 89% of Americans rated a commitment to accuracy as very important for trusting a news organization, and 86% said the same of correcting errors quickly and publicly.
What makes a reference solid is its permanent identifier. For an article, that is the DOI: a fixed address given to the publication, which keeps working even after the journal rebuilds its website. For a book, it is the ISBN. For you, the author, it is the ORCID: a unique number that gathers your work in one place and stops anyone confusing you with a namesake. All three codes exist to remove doubt. A degree says what you were taught; a publication says what you have produced since, and when. On a physiotherapist's page, that looks like one line: the exact article title, the journal name, the month and year, then the clickable DOI that opens the official page. On the clinic sites I audit, the founder's publications page is often the oldest page on the site: titles copied from a 2016 CV, not one link, when half the articles have a DOI that only needs pasting.
On sensitive topics, a badly cited publication costs more than no publication at all. If your site covers health, money, law or safety (Google's label for those topics is YMYL, short for Your Money or Your Life), the expected level of proof goes up. Announcing "published in international journals" without a single openable reference reads as an inflated claim, and that is exactly what an evaluator is trained to spot. On the AI side the logic is mechanical. When ChatGPT, Perplexity or Google's AI summaries pick who to cite, they look for facts they can confirm somewhere else. An article with a DOI exists in public databases such as Crossref or PubMed, so a machine can confirm it is real. A title with no identifier exists only on your page, and carries no weight.
The NeuroEEAT scoring grid (0-10)
Rate what a stranger can verify about your publications in three minutes, without taking your word for it.
| Score | Description | Concrete indicators |
|---|---|---|
| 0-2 | Nothing citable | No publication mentioned at all, or a vague line such as "author of numerous articles". No title, no date, no journal, no link. |
| 3-4 | Titles and nothing else | A list of titles with no date and no publisher, or links that lead nowhere any more. The reader has to go hunting to find out whether the work exists. |
| 5-6 | Decent references | Exact title, journal and year on most lines, but no permanent identifier and no clear connection to the author who signs the site. |
| 7-8 | Verifiable references | Every publication has its exact title, its journal or publisher, its date and a clickable DOI or ISBN. Co-authors are named. The list carries a last-updated date. |
| 9-10 | Proof tied to the content | Publications are attached to a single author profile (ORCID or equivalent), cited inside the articles they genuinely support, and declared in the page code (JSON-LD, the data format Google reads). Nothing sits in the list that cannot be opened. |
How to audit it on your site
Open your publications page and test it the way a stranger would. Allow thirty minutes.
- 01
Open every link on your list
One by one, in a private browsing window. A dead link, a missing PDF, a paywall that does not even show the abstract: each of those drops the reference back to being a bare title. Count how many of your publications survive the test.
- 02
Check that every line is complete
Four elements are needed: the exact title, the name of the journal or publisher, the date, and the identifier (DOI, ISBN or official link). If one is missing, the line needs fixing. The date is the one most often left out, and it is the one that tells a reader whether the work still counts.
- 03
Search your own name in public databases
Type your name into Google Scholar, Crossref or PubMed, depending on your field. What you find there is what an evaluator or an AI tool will find too. If your site claims ten publications and nothing comes up, the gap is visible to everyone.
- 04
Compare the byline on the site with the one in the journal
The name on your author page has to be written exactly as it appears in the publication. "J. Bedard" in one place and "Jacob Bedard" in another, on two different sites, breaks the connection for a machine. Check the co-authors are listed too: dropping them is spotted in seconds.
- 05
Connect each publication to a page on the site
A publication earns its keep when it backs up a specific piece of content. Find the articles on your site that cover the same subject and never mention it. That is where the proof is missing, not on the bio page.
How to activate or improve it
From a first clean list to references that keep themselves up to date.
Build a list people can open
- Gather every publication into a single document, the old ones included.
- For each, write down four things: exact title, journal or publisher, month and year, official link.
- Add the DOI when there is one: it is printed on the first page of the article, or shown on the journal page.
- Publish the list on your author page, newest first.
- Delete vague lines such as "author of numerous articles": they weaken everything that follows.
Make the proof verifiable
- Create an ORCID, free and done in ten minutes, attach your publications to it and link it from your bio.
- Write your name the same way everywhere: site, journal, professional profiles, social accounts.
- Name every co-author, in the order used in the original publication.
- Cite the right publication inside the articles that cover the same subject, with one sentence saying what it adds.
- Date the list and reopen every link twice a year.
Turn it into a system
- Declare your publications in the page code (JSON-LD, using ScholarlyArticle or Book types attached to your Person profile) so machines read the same list as your visitors.
- Deposit the accepted version in an open archive such as arXiv or your institution's repository when the rights allow it, and say which version you are hosting.
- Keep one page per major publication, with a plain-language summary of what you found and who it helps.
- Tie your publications to your other proof: degree, certification, speaking slots, published datasets.
- Check once a quarter who cites you: citations received are authority evidence you did not write yourself.
5 common mistakes to avoid
Sector specifics
The bar for proof goes up
On a health topic, a publication is the strongest expertise signal you have, but it still has to be readable by a patient. Give the full reference, then sum up in two sentences what the study showed and what it did not. Never turn work done on a small group into a general promise: that is precisely what an evaluator flags as risky information.
Trade publishing beats academic silence
Few lawyers and advisers publish in peer-reviewed journals, and that is fine. An article in a legal review, a chapter in a professional handbook or a technical note from a recognised publisher does the same job. The rule does not change: publisher, date, link or ISBN. Just check the piece respects your duty of confidentiality about client matters.
The trade magazine counts
A roofer or a baker does not publish studies, but an article in a trade magazine, a technical guide from a trade body or a regional professional publication are publications in full. Cite the name of the outlet and the date. If the text does not exist online, scan the page and host the file yourself, stating the issue number and month.
A white paper is not a publication
A white paper you publish yourself does not belong in this signal: nobody outside validated it. It stays useful, but keep it separate, in your resources, away from references that went through a publisher. Mixing the two casts doubt on the real ones, which is the opposite of the point.
Signals to activate in parallel
A publication on its own is a line in a list. These signals decide whether it is believed:
FAQ
How to cite your publications when you only have two?
Cite them in full and stop there. Two references a reader can open beat a sentence that hints at ten. Put them on your author page with title, journal, date and link, then bring the more relevant one back into the articles on your site that cover the same subject.
Do you need a scientific paper to score well?
No. The signal recognises three forms: the scientific article, the book with a publisher, and the professional publication (trade magazine, technical handbook, note from a recognised body). What matters is that somebody outside agreed to publish your text and that it can be found again. With none of the three, lean on the other expertise signals instead.
What do you do when your article sits behind a paywall?
Cite it anyway, with its DOI. The official page shows at least the title, the authors, the journal and the abstract, which is enough to prove it exists. If the journal allows it, deposit the accepted version in an open archive and give both links. Never re-upload the final PDF to your own site without checking the rights first.
Does a publication from 2014 still count?
Yes, but not in the same way. It proves a track record, not that you are current. Keep it, with its date, and add something recent even if it is small: a technical note, a conference talk, a published dataset. A list that stops ten years ago tells the reader something, and it is not what you want it to say.
Where should the list live on the site?
On the author page, and nowhere else in full. Individual articles then cite only the publications useful to their subject, with a link back to the author page. Repeating the whole list at the bottom of every page tires the reader and adds no signal at all.
Case study
Before · a list nobody can open
After · a list that checks out
Illustrative figures meant to show the typical progression of the signal. They do not come from a real client case and must be replaced with your own measurements before any external communication.