Short answer: if a publisher is inconsistently represented in Search or AI responses, don't start by rewriting articles. Check editorial brand, canonical domain, author profiles, aliases, syndication and topic owners. Google documents Organization, ProfilePage and Article structured data, but does not publish an entity salience score. The diagnosis must find concrete conflicts.
Failure mode 1: editorial brand versus parent company
The publisher may belong to a media group with a different name. If the About page, Organization markup, and external profiles use the entities interchangeably, the system becomes ambiguous.
Define which name describes the publication and which name describes the legal owner.
Failure mode 2: site name unstable
Migration or rebranding may leave the old name in titles, metadata, logo alt text and external profiles. A mix of current and historical names should not be ignored.
Failure mode 3: duplicate author profiles
An old and a new CMS can create separate URLs for the same person. Bylines are shared between profiles, and external references can use any of them.
Failure mode 4: aliases not registered
An author may publish under a professional name different from the legal one. Legitimate pseudonym should be documented, not automatically "corrected".
Failure mode 5: unclear syndication
Articles republished on partners can create multiple versions. Entity salience does not resolve URL custody. You need syndication mapping and editorial rules.
Failure mode 6: topic ownership unclear
Multiple articles may define the same concept. Clean entity identity does not decide which page is the owner. The audit must include duplicate intents.
Failure mode 7: bad external profiles
A company profile can describe the media group, not the publication. An author profile may belong to another person with a similar name.
Verify identity before `sameAs' or linking.
Failure mode 8: co-authors lost on import
Migration can only keep the first author. This is an issue of attribution and integrity, not just schema.
Failure mode 9: contradictory article markup
Visible byline says one author, JSON-LD another. It fixes the mismatch, but does not promise an external outcome.
Failure mode 10: section brands treated as separate publications
Some sections have their own visual identity but are not independent entities. Do not create separate Organization markup without real basis.
Failure mode 11: rewrite precedes audit
The team notices a decrease in citations and rewrites the introductions, although the real problem is canonical conflict or duplicate profiles. The intervention is on the wrong layer.
Failure mode 12: an "entity score" hides P0
An 88/100 score can look good even if the profiles of two authors are reversed. Material findings should be shown separately from any internal aggregation.
Reproducible decision tree
- Is the publishing brand clearly separated from the parent company?
- Is the site name stable?
- Is the canonical domain clear?
- Are author profiles unique?
- Are aliases and pseudonyms documented?
- Do Byline and Article markup match?
- Are co-authors retained?
- Is the syndication map known?
- Do external profiles really describe the same entity?
- Are topic owners clear?
- Is duplicate intent under control?
- Can Findings be rechecked after remediation?
If 1-8 have problems, rewriting the copy is not the first action.
How you build the audit
Export organization/site metadata, author registry, profile URLs, bylines, markup, canonical URLs, aliases and syndication relationships. Add topic owner for priority pages.
Stores raw values alongside the verdict.
Severity
P0: wrong author or brand identity, wrong canonical domain. P1: duplicate profiles, syndication conflict, lost co-authors. P2: aliases and external profiles stale. P3: cosmetics.
How to check after fixing
Re-crawl and compare the same URLs. For author merges, check a sample of historical articles, not just new profiles.
Finding is closed when the corrected surface is observable, not at commit.
What you don't measure here
Does not include ranking, citations and revenue in entity health. These are external outcomes and must be monitored separately.
Example of correct diagnosis
A publication is cited under the old name after the rebrand. The audit shows that the homepage is correct, but 40% of author profiles and the About page in EN use the old name. The fix is identity cleanup and alias mapping, not rewriting hundreds of articles.
Stop criterion
Switch to monitoring when brand, domain and authors are consistently identified, P0/P1 are zero and new imports do not reintroduce duplicates. Don't keep going just to increase a score.
How do you deal with editions and sub-brands
Some publications have self-titled sections, podcasts or verticals that seem like independent brands. In audit, classify publication, section, product and series before comparing profiles. A social account for a podcast is not automatically `sameAs' for the main publishing organization.
How to handle author profiles in multiple languages
An author can have RO and EN profiles with the same identity. Do not consider them duplicates if each serves a localized edition and canonical mapping is intended. Finding occurs when bylines send inconsistently without a rule.
Re-audit threshold
Resume auditing after CMS migration, rebrand or author structure change. Otherwise, it monitors duplicate profiles, aliases and canonical conflicts incrementally.
Claim ledger
- FACT/EVIDENCE: Google documents Organization, ProfilePage and Article structured data.
- PRACTITIONER GUIDANCE: entity diagnostics for publishers must include authorship, aliases, canonical and syndication.
- INFERENCE: reducing conflicts can reduce naming ambiguity.
- NOT PROVEN: a universal entity salience score or guaranteed effect on AI ranking/citations.
Conclusion
Entity salience is useful as a diagnostic only if you translate it into verifiable conflicts. For publishers, brand, authors and syndication are more important than an abstract score. Fix the identity before rewriting the corpus.
Sources reviewed
- Google Search Central, Organization structured data: https://developers.google.com/search/docs/appearance/structured-data/organization
- Google Search Central, ProfilePage structured data: https://developers.google.com/search/docs/appearance/structured-data/profile-page
- Google Search Central, Article structured data: https://developers.google.com/search/docs/appearance/structured-data/article
