What Are Trust Signals for AI Search?
2.4x
Citation rate lift for pages with Person schema
Versus equivalent pages without author attribution
70.4%
of ChatGPT citations include Person schema
EverTune analysis of 50,000 ChatGPT-cited URLs
Trust signals are the machine-readable equivalent of a professional credential — they allow AI to assess whether a page is safe to cite without reading every word. The 10 categories below are ordered by impact. Implement them starting with author attribution, which provides the highest return per hour invested.
Author Attribution (HIGH Impact)
HIGH IMPACTMinimum viable Person schema
{
"@context": "https://schema.org",
"@type": "Person",
"name": "[Author Full Name]",
"url": "[Author Bio Page URL]",
"jobTitle": "[Relevant Professional Title]"
}YMYL-enhanced Person schema (health, finance, legal pages)
{
"@context": "https://schema.org",
"@type": "Person",
"name": "[Author Full Name]",
"url": "[Author Bio Page URL]",
"jobTitle": "[Credential-Bearing Title]",
"honorificSuffix": "[MD / PhD / JD / CFA]",
"hasCredential": {
"@type": "EducationalOccupationalCredential",
"credentialCategory": "[Medical License / Bar Admission / CFA Charter]",
"recognizedBy": {
"@type": "Organization",
"name": "[Licensing Body]"
}
},
"sameAs": [
"[LinkedIn Profile URL]",
"[ORCID URL or Twitter URL]"
]
}Byline detection locations (AI checks in order)
- 1Person schema JSON-LD with name field
- 2Visible byline element with author-role CSS class or HTML attribute near headline
- 3About the Author section within article body
- 4Visible text matching “By [Name]” or “Author: [Name]” within first 500 words
Missing all four locations → “author absent” flag.
About Page Quality
The About page is the second most-checked trust signal after author attribution. AI systems access the /about URL directly when evaluating site-level credibility. A missing or thin About page is a trust signal failure that affects every page on the domain. **Minimum requirements.** The About page must exist at /about (or /about-us) and contain at least 200 words of substantive content. The 200-word floor corresponds to AI systems’ thin-content threshold — below this count, the page may be classified as having insufficient credibility signal. **What AI checks on your About page:** 1. **Existence check:** Does /about exist and return a 200 HTTP status? A 404 or redirect is a trust failure. 2. **Named individuals:** Are there named humans associated with the organization? A company-only About page with no people is weaker than one with named founders, team members, or leadership. 3. **Credential language:** Does the page contain language about relevant expertise, years of experience, certifications, or industry recognition? 4. **Organization details:** Does the page include founding year, location, or mission statement? These details help AI verify organizational legitimacy. 5. **Schema presence:** Is Organization schema present with name, url, logo, and description fields? **High-impact About page additions:** - Named founders or leadership team with titles - Founding year and organization history - Mission or values statement - Relevant credentials, certifications, or awards - Links to press mentions or media coverage - Organization schema with sameAs links to social profiles **The “who is behind this site” test.** AI systems effectively apply a “who is behind this site” evaluation when assessing YMYL or commercial content. If a user could not determine who runs the site from the About page alone, the trust signal is insufficient.
Contact Information
Contact information is a baseline trust signal that confirms organizational legitimacy. AI systems and search engines both use the presence of accessible, verified contact information as a proxy for organizational accountability. **What contact information AI looks for:** - Email address (text or mailto: link in the HTML — not an image, not behind a form-only wall) - Physical address (street address for physical businesses; at minimum a city and country for remote organizations) - Phone number (recommended for e-commerce, medical, legal, and financial sites) - A functional contact form at /contact **The footer link standard.** A link to /contact in the site footer is the minimum acceptable implementation. The contact page itself must exist (not 404), must return a 200 HTTP status, and must contain at least one of the above contact methods. **ContactPoint schema.** Adding ContactPoint schema to your Organization schema explicitly makes contact information machine-readable: { "contactPoint": { "@type": "ContactPoint", "contactType": "customer service", "email": "support@yourdomain.com", "availableLanguage": "English" } } **Why contact information matters for AI.** AI systems are less likely to cite a source that provides no way for users to follow up, report errors, or verify information. A page with accessible contact information signals that the organization is accountable for what it publishes.
Legal Pages and Disclaimers
Legal pages (privacy policy, terms of service) are trust signals that confirm a site operates within standard legal frameworks. Their absence is a minor trust signal failure for general content; it is a significant failure for YMYL content. **Required legal pages:** - Privacy Policy: must exist at /privacy-policy or /privacy; must address data collection practices; must be accessible via footer link - Terms of Service: required for e-commerce, SaaS, and any site where users create accounts or submit data **YMYL escalation.** For health, medical, financial, investment, legal, and safety content, legal pages alone are insufficient. YMYL pages require additional disclaimers: *Medical content disclaimer:* “This content is for informational purposes only and does not constitute medical advice. Consult a qualified healthcare provider before making medical decisions.” *Financial content disclaimer:* “This content is for educational purposes only and does not constitute financial advice. Past performance does not guarantee future results. Consult a licensed financial advisor before making investment decisions.” *Legal content disclaimer:* “This content provides general legal information and does not constitute legal advice. For advice specific to your situation, consult a licensed attorney in your jurisdiction.” **Placement requirements.** Disclaimers must appear within the article body — not only in a footer or sidebar. AI systems evaluate disclaimer presence relative to the YMYL content, so a page-level disclaimer that is part of the article content is stronger than a site-wide footer disclaimer. **Schema for legal pages.** Adding WebPage schema with a “legal information” type to privacy policy and terms of service pages makes them machine-readable as legal documents, which strengthens the site-level trust profile.
Publication and Update Dates
Publication and update dates are verifiable freshness signals that AI systems use both to assess content currency and to evaluate publishing discipline. A page with clear, consistent dates signals that a real organization is maintaining the content. **Required date signals:** - datePublished in Article schema JSON-LD (ISO 8601 format: YYYY-MM-DD) - dateModified in Article schema JSON-LD (update whenever content changes) - Visible publication date near the article headline (human-readable format) - Visible “last updated” date when content has been substantially revised **The consistency check.** AI systems compare the datePublished in schema against the visible date shown on the page. If the schema date is significantly different from the visible date — for example, schema says 2024-01-15 but the page shows “Published March 2023” — this inconsistency is a trust signal failure. Schema dates and visible dates must match. **The artificial freshness trap.** Updating dateModified without making substantive content changes is detectable. AI systems and search quality evaluators look for whether the dateModified update is accompanied by actual content changes. A page that updates its dateModified weekly without changing any content will eventually be flagged for artificial freshness manipulation. **Best practices:** - Display publication date in a standard location (below headline, above byline, or below byline) - Update dateModified and the visible date simultaneously when making changes - For evergreen content, add a visible “Reviewed [month, year]” note when you review but do not substantially change content - Never delete publication dates to hide age — this removes a trust signal rather than improving it
External Citation Quality
34.9%
Pages with .edu/.gov citation
3.2%
Pages with no citations
10.9x
AI selection lift
Zyppy citation pattern research
.edu domains
University research centers, academic studies
.gov domains
CDC, NIH, FDA, SEC, FTC, Bureau of Labor Statistics
Peer-reviewed journals
PubMed, Nature, NEJM, JAMA, Science
Major research firms
Gartner, McKinsey, Forrester, Pew Research, IDC
Standards bodies
IEEE, ISO, ANSI, W3C, NIST
Official documentation
Google Search Central, AWS docs, platform APIs
First-Hand Experience Signals
First-hand experience is the E in E-E-A-T. AI systems preferentially cite content containing signals they cannot generate themselves: specific numeric outcomes, failure accounts, iteration details, and proprietary observations.
- • Specific numeric outcomes: “bounce rate dropped from 74% to 51% in 6 weeks”
- • Named examples: “when we ran this test on 47 URLs across three client sites”
- • Failure accounts: “the first approach did not work because...”
- • Time-based observations: “over 90 days of monitoring, we found”
- • Generic advice: “many website owners find that...”
- • Only public data with no proprietary analysis
- • No failure or limitation acknowledgment
- • Content that mirrors structure of existing top results
YMYL Classification and Escalation
YMYL content is held to a higher E-E-A-T standard because inaccurate information in these categories can cause real-world harm. AI systems are disproportionately cautious about citing YMYL pages that lack the required signals below.
| Category | Required Credentials | Citation Sources | Disclaimer |
|---|---|---|---|
| Health & Medical | MD, DO, NP, PA, RN, PharmD — or content reviewed by a named credentialed professional | PubMed, Cochrane, CDC, NIH, NICE, official clinical guidelines | Medical disclaimer within article body. MedicalWebPage schema recommended. |
| Finance & Investment | CFA, CFP, CPA, or licensed financial advisor reviewer named explicitly | SEC, FDIC, Federal Reserve, Bureau of Labor Statistics, recognized financial data providers | Financial disclaimer + ‘not investment advice’ statement within article body. |
| Legal | JD or bar admission in the relevant jurisdiction, named as author or explicit reviewer | Specific statutes, regulations, or case law by official citation | Legal disclaimer + jurisdiction-specific caveat within article body. |
| Safety | Professional safety certification in the relevant domain (OSHA-30, licensed electrician, etc.) | OSHA standards, NFPA codes, equivalent safety standards body references | Safety warning at the beginning of any hazardous procedure instructions. |
Trust Signal Checklist
Required for AI citation. Missing any of these significantly reduces citation likelihood.
Strongly increase AI citation likelihood. Implement after Critical items are complete.
Contribute to overall trust evaluation. Prioritize for competitive or YMYL content.
AI Content Disclosure
Google’s 2025 guidance does not require AI disclosure. It requires that all content — regardless of authorship tool — meet the same helpfulness and E-E-A-T standards. For YMYL content, disclosure of AI involvement prompts additional E-E-A-T scrutiny.
Disclosure template (recommended language)
This article was [written / drafted] with the assistance of AI writing tools and reviewed by [Author Name], [Credential], for accuracy and editorial standards. The research, original analysis, and editorial decisions are the work of [Author/Organization Name].
Place immediately below the article headline or within the author byline section. Footer-only disclosure is weaker than in-article disclosure.
Content patterns that trigger AI-content flags (without disclosure)
- • Perfect structural alignment with other AI-generated pages on the same topic
- • Absence of any first-hand experience signals
- • Uniform paragraph length and identical heading patterns across multiple pages
- • Hallucinated or non-verifiable statistics without source links
- • Hedging language patterns characteristic of large language model output
Audit Your AI Search Visibility
See exactly how AI systems view your content and what to fix. Join the waitlist to get early access.
Frequently Asked Questions
Yes, for optimal trust signal value. AI systems are specifically looking for attributable, verifiable authorship. A named person whose credentials can be cross-referenced via sameAs links (LinkedIn, ORCID, institutional profile) provides a stronger trust signal than a pseudonym or collective byline. Using a pen name is not prohibited, but it eliminates the cross-referencing benefit of Person schema.
Organization authorship is weaker than individual Person authorship for AI trust signals. A company byline (“By Acme Corp Editorial Team”) provides no credential information and cannot be cross-referenced. For content where trust matters — especially YMYL content — a named individual author is strongly recommended. For marketing copy and commercial pages, organization authorship is acceptable because those page types are not evaluated on E-E-A-T to the same degree.
The easiest path: install a schema plugin such as Rank Math or Yoast SEO Premium, which generate Person schema from author profile fields. Fill in the author’s name, title, bio, and social profile URLs in the author profile settings. The plugin will generate the JSON-LD automatically on any post assigned to that author. Alternatively, add the JSON-LD manually to your child theme’s functions.php using wp_head hook, or use a header/footer injection plugin to add the schema to specific post types.
There is no published minimum threshold — citation selection is probabilistic, not pass/fail. However, from citation pattern research: pages with Person schema are cited at 2.4x the rate of pages without it; pages with external citations are cited at up to 10.9x the rate of pages with none; pages with datePublished and dateModified are cited at higher rates than pages with no date signals. The combination of all critical trust signals (author attribution, dates, external citations) dramatically increases citation probability.
No. E-E-A-T requirements scale with YMYL classification. Informational content on non-YMYL topics (how to organize a closet) requires minimal E-E-A-T beyond basic author attribution. Commercial content (software comparison pages) requires moderate E-E-A-T including author attribution and organization legitimacy signals. YMYL content (medical, financial, legal) requires maximum E-E-A-T: named credentialed author, peer-reviewed citations, and topic-specific disclaimers. Apply effort proportionally to the YMYL level of your content.
Place the disclosure within the article body, not only in a footer. Recommended language: ‘This article was drafted with AI writing assistance and reviewed by [Author Name], [Credential], for accuracy.’ Specify the nature of AI involvement (drafted, outlined, researched) and the nature of human involvement (reviewed, edited, added original analysis). Avoid vague disclosures like ‘AI-assisted’ with no additional context — specific disclosures are more trustworthy than generic ones.
A Wikipedia page is a significant trust signal — but only if it exists organically. Wikipedia has strict notability requirements: a subject must have significant coverage in reliable, independent sources. If your organization or key individuals have Wikipedia pages, including the Wikipedia URL in Person or Organization schema sameAs fields provides a high-authority cross-reference. However, creating a Wikipedia page solely for SEO or trust purposes violates Wikipedia policy and will be deleted.
Yes, substantially. The highest-impact trust signal improvements — Person schema, visible publication dates, external citation links, legal page existence, contact information — can all be implemented without altering the main article body. Adding a byline, adding schema JSON-LD to the page head, linking to existing authoritative sources, and creating or improving the /about and /contact pages are all structural changes that do not require rewriting content.
Audit Your AI Search Visibility
See exactly how AI systems view your content and what to fix. Join the waitlist to get early access.
Audit Your AI Search Visibility
See exactly how AI systems view your content and what to fix. Join the waitlist to get early access.