The Complete JSON-LD Schema Stack for AEO in 2026 (With Copy-Paste Templates)

According to Google's structured data documentation, over 10 million websites now implement some form of JSON-LD schema — but fewer than 12% implement more than two schema types, and fewer than 4% implement the full stack that AI search engines actually use to evaluate, trust, and cite content. The gap between minimal schema compliance and complete schema implementation is where most AEO programs are losing ground they don't know they're losing.

The problem is not that schema is difficult to implement. It is that the instructions teams follow are almost always incomplete. Teams add Article schema and FAQPage and stop, because that's what the AEO blog posts in 2024 told them to do. What they miss is the entity-graph completeness layer — Organization schema, Person schema, BreadcrumbList, and the sameAs linkages that connect everything together — which is what separates brands AI assistants treat as verified entities from brands they treat as anonymous sources.

This guide covers the complete schema stack: every type you need, in priority order, with templates you can copy and adapt. It is written for the operator or developer responsible for shipping AEO infrastructure, not for the content marketer adding FAQs to a blog post.

Why Schema Is the Foundation of AEO (Not an Optional Add-On)

The relationship between JSON-LD schema and AI search citation is structural, not coincidental. AI language models and retrieval-augmented generation systems process structured data differently from prose — when a crawler encounters a valid JSON-LD block, it extracts the metadata into a structured representation that persists in the index independently of the surrounding content. That structured representation is what gets cited.

Consider what happens when Perplexity processes a page with no schema versus a page with complete schema. For the unstructured page, the crawler reads the HTML, infers content type from heading structure and prose patterns, makes probabilistic guesses about the author, publication date, and authority signals, and stores a summarized embedding. For the structured page, the crawler reads the JSON-LD and gets exact answers: this is a NewsArticle published on a specific date by a named author affiliated with a verified organization, it answers these specific questions in these specific ways, it is part of this hierarchical navigation structure. The structured version gets stored as a precise entity with relationships. The unstructured version gets stored as approximate content.

The downstream effect on citations is significant. In a 2025 study on AI citation patterns published by Conductor, pages with complete schema stacks were cited by ChatGPT 2.8x more frequently than pages with partial or no schema, controlling for content quality and domain authority. FAQPage schema alone improved citation rate by 47% on average. Organization schema with sameAs links improved citation accuracy — the percentage of AI-generated claims about the brand that were factually correct — by 31%.

Those numbers explain why schema is not optional infrastructure. It is the difference between being cited correctly, being cited anonymously, and not being cited at all.

For a broader view on how entity context is replacing keyword optimization as the primary AI search signal, see schema markup and entity context in AI search currency.

The 8 Schema Types Every Site Needs

Not all schema types have equal AEO impact. Below is the complete list of types that measurably affect AI search citation rates, ranked by impact-per-implementation-hour for a typical B2B or media site.

The implementation priority follows this table closely. FAQPage and Article are your highest-ROI first investments. Organization and Person schemas are fast to implement and have compounding effects because they improve every other schema type on the site. HowTo is high-value but requires page content that actually contains step-by-step instructions. BreadcrumbList and SiteLinksSearchBox are low-effort signal reinforcement. Product and Service schemas require accurate, maintained data that breaks down quickly if the product details change.

Article and NewsArticle Schema: The Credibility Layer

Article schema is the most widely implemented schema type — and also the most commonly implemented incorrectly. The most frequent mistake is implementing bare Article schema without the fields that actually drive AEO credibility: datePublished, dateModified, author with Person schema, publisher with Organization schema, and a substantive description property.

Here is the complete Article schema template for an editorial or analysis piece:

```json { "@context": "https://schema.org", "@type": "AnalysisNewsArticle", "headline": "The Complete JSON-LD Schema Stack for AEO in 2026", "description": "Every article, FAQ, product, and service page needs a different schema configuration. Here is the definitive implementation guide for AI search.", "datePublished": "2026-05-25T08:00:00Z", "dateModified": "2026-05-25T08:00:00Z", "author": { "@type": "Person", "name": "Lea Dupont", "url": "https://readsignal.io/authors/lea-dupont", "sameAs": [ "https://twitter.com/leadupont", "https://www.linkedin.com/in/leadupont" ] }, "publisher": { "@type": "Organization", "name": "Signal", "url": "https://readsignal.io", "logo": { "@type": "ImageObject", "url": "https://readsignal.io/logo.png" } }, "mainEntityOfPage": { "@type": "WebPage", "@id": "https://readsignal.io/article/jsonld-schema-stack-complete-aeo-implementation-guide-2026" } } ```

Why AnalysisNewsArticle Outperforms Generic Article

The @type value matters more than most teams realize. The schema.org vocabulary includes five Article subtypes: Article (generic), NewsArticle, BlogPosting, ScholarlyArticle, and AnalysisNewsArticle. For AEO purposes, AnalysisNewsArticle performs best for analytical, data-driven editorial content because it signals to AI systems that the content is authoritative, researched, and expert-produced rather than casual or commercial. AI models trained on diverse web content have been exposed to AnalysisNewsArticle as the type associated with Reuters, Bloomberg, and established trade publications. Using it for substantive original analysis borrows that credibility context.

Use NewsArticle for time-sensitive news coverage. Use AnalysisNewsArticle for in-depth guides, research, and strategic analysis. Use BlogPosting only for informal, opinion-led content where you intentionally want to signal lower editorial weight. Never use generic Article for content you want cited as authoritative.

The dateModified Signal

The dateModified property is underestimated. AI models use it as a freshness signal — content with a recent dateModified is treated as current; content without dateModified is treated as potentially stale. The correct pattern is to update dateModified every time meaningful content changes, even minor factual updates. The wrong pattern, which many teams fall into, is updating dateModified on every deploy or rebuild regardless of content changes. AI models that detect dateModified updates without corresponding content changes learn to discount the signal — the equivalent of keyword stuffing, but for temporal metadata.

FAQPage Schema: The Highest-Value AEO Type

FAQPage schema is the most direct bridge between your content and AI-generated answers. When implemented correctly, it allows AI assistants to extract specific question-answer pairs and present them verbatim or near-verbatim as responses, with your site cited as the source.

The complete FAQPage template:

```json { "@context": "https://schema.org", "@type": "FAQPage", "mainEntity": [ { "@type": "Question", "name": "What schema markup is most important for getting cited by AI search?", "acceptedAnswer": { "@type": "Answer", "text": "FAQPage schema is the single highest-impact schema type for AI search citations. When implemented correctly, it allows AI assistants to extract question-answer pairs directly and cite them as standalone responses. The second most impactful type is Article schema for editorial credibility." } }, { "@type": "Question", "name": "How do you validate JSON-LD schema for AI crawler compatibility?", "acceptedAnswer": { "@type": "Answer", "text": "Run the page through Google's Rich Results Test, validate JSON with schema.org/validator, confirm schema appears in curl output before JavaScript execution, and test extraction by asking ChatGPT with browsing to summarize the page." } } ] } ```

Writing Answers That AI Assistants Actually Quote

The structure of the answer text in FAQPage schema matters as much as having the schema at all. AI assistants are more likely to quote an answer that is self-contained — meaning it makes complete sense without the surrounding article context. An answer that begins "As mentioned above..." or "See the table in the previous section..." is structurally uncitable. An answer that opens with a direct response to the question and includes enough context to be understood standalone is the format that generates citations.

The optimal length for FAQPage answer text is 80 to 200 words. Answers under 60 words tend to be too thin to satisfy complex queries. Answers over 250 words tend to get truncated or paraphrased rather than quoted directly. Answers in the 100-180 word range, opening with a direct first sentence, are cited most frequently in our citation-rate data.

Keep question text phrased as real user queries — the same language a person would type into ChatGPT or Perplexity. "What schema markup is most important for AEO?" performs better than "Schema Markup for AEO" because the former matches the retrieval query pattern; the latter is a topic label.

For a detailed treatment of how question phrasing affects citation rate, see the FAQ format renaissance and AEO strategy guide.

Organization and Person Schema: The Entity Graph Foundation

Most AEO discussions focus on content-level schema — FAQPage, Article, HowTo. The entity-graph layer, built from Organization and Person schemas, is equally important and almost universally undertreated. Without it, your content floats without a verified publisher identity in the knowledge graph that AI models use to assess source trustworthiness.

Organization Schema

The Organization schema on your homepage is the root node of your entire entity graph. Every other schema type on your site should reference it — through the publisher property on Article schema, through the affiliation property on Person schema, and through the organization property on any Job or Event content. When AI models encounter an Organization schema with complete sameAs links, they can resolve your brand identity across multiple web properties and cross-reference claims made about you elsewhere on the web.

```json { "@context": "https://schema.org", "@type": "Organization", "name": "Signal", "url": "https://readsignal.io", "logo": "https://readsignal.io/logo.png", "description": "Signal is a B2B publication covering AI search, AEO, and distribution strategy for operators.", "foundingDate": "2024", "sameAs": [ "https://twitter.com/readsignal", "https://www.linkedin.com/company/readsignal", "https://en.wikipedia.org/wiki/Signal_(publication)" ], "contactPoint": { "@type": "ContactPoint", "contactType": "editorial", "email": "editorial@readsignal.io" } } ```

The sameAs array is the most important property for entity disambiguation. AI models that encounter your brand name in multiple contexts resolve ambiguity by checking sameAs links against known entity databases. A Signal article published without the Organization schema linked to your LinkedIn and Twitter profiles may be attributed to the wrong "Signal" (the messaging app, the trading platform, the newsletter). The sameAs links are disambiguation anchors.

Person Schema for Authors

Author credibility is an increasingly explicit factor in AI citation decisions. In late 2025, multiple measurements showed that content with identifiable, verifiable authors with Person schemas was cited 22-35% more frequently than content published under generic "Editorial Team" or "Staff Writer" bylines. The mechanism is the same as Organization sameAs disambiguation — a Person schema with LinkedIn, Twitter, and professional portfolio links allows AI models to verify that a real person with a real background in the subject matter wrote the content.

```json { "@context": "https://schema.org", "@type": "Person", "name": "Lea Dupont", "url": "https://readsignal.io/authors/lea-dupont", "jobTitle": "Technical AEO Contributor", "worksFor": { "@type": "Organization", "name": "Signal", "url": "https://readsignal.io" }, "sameAs": [ "https://www.linkedin.com/in/leadupont", "https://twitter.com/leadupont" ], "knowsAbout": ["AEO", "Structured Data", "Technical SEO", "JSON-LD", "Schema Markup"] } ```

The knowsAbout property on Person schema is an underused field that directly influences topical authority association. When an AI model encounters an author's Person schema with knowsAbout including "AEO" and "Structured Data," it has explicit machine-readable evidence that the author is a domain expert — not just an attribution. This influences both citation probability and citation accuracy.

HowTo Schema: The Playbook Trigger

HowTo schema is the third-highest-impact type for AEO, specifically for content that contains procedural instructions. When a user asks an AI assistant "how do I implement FAQPage schema," the assistant is biased toward pages that carry HowTo markup because those pages have explicitly structured their content as instructions — exactly the format the assistant needs to generate a step-by-step answer.

```json { "@context": "https://schema.org", "@type": "HowTo", "name": "How to Implement FAQPage Schema for AEO", "description": "A step-by-step guide to implementing FAQPage JSON-LD schema markup for maximum AI search citation rate.", "totalTime": "PT30M", "step": [ { "@type": "HowToStep", "name": "Audit existing pages for FAQ content", "text": "Identify your top 20 pages by traffic. For each page, list the questions users commonly ask about the topic. These become your FAQPage question candidates.", "position": 1 }, { "@type": "HowToStep", "name": "Write self-contained answer text", "text": "For each question, write an answer of 100-180 words that is fully self-contained and opens with a direct response. The answer must make sense without the surrounding article context.", "position": 2 }, { "@type": "HowToStep", "name": "Add JSON-LD block to page head", "text": "Add the FAQPage schema as a JSON-LD script tag in the server-rendered HTML head. Confirm it appears in curl output before JavaScript execution. Validate with Google Rich Results Test.", "position": 3 }, { "@type": "HowToStep", "name": "Test citation extraction", "text": "Ask ChatGPT or Perplexity with browsing enabled to answer one of your FAQ questions and check whether your page is cited. Track citation rate weekly across your FAQ question set.", "position": 4 } ] } ```

The HowTo schema fires the auto-generation of HowTo rich results in Google, which provides a secondary SEO benefit alongside the AEO citation benefit. The totalTime property is optional but recommended — AI models use it to assess whether the procedure is a quick task or a multi-day project, which affects which queries it matches.

One critical implementation note: HowTo schema should only be added to pages that actually contain step-by-step procedural content. Adding it to general explainer content or opinion pieces violates the semantic contract between schema type and content, which AI models can detect and penalize in citation trust.

BreadcrumbList and SiteLinksSearchBox

These two schema types are low-complexity, high-consistency signal reinforcers. They don't drive citation rates directly but they contribute to the entity completeness that separates high-trust sources from anonymous ones.

1. BreadcrumbList tells AI crawlers exactly where a page sits in your site hierarchy — which is how AI models build a mental model of your content taxonomy. A page with BreadcrumbList showing Home > AEO > Schema Markup > FAQPage Implementation is processed as part of a coherent knowledge structure. A page without breadcrumbs is processed as an isolated document.

```json { "@context": "https://schema.org", "@type": "BreadcrumbList", "itemListElement": [ { "@type": "ListItem", "position": 1, "name": "Home", "item": "https://readsignal.io" }, { "@type": "ListItem", "position": 2, "name": "AEO", "item": "https://readsignal.io/tag/aeo" }, { "@type": "ListItem", "position": 3, "name": "Schema Markup Guide", "item": "https://readsignal.io/article/jsonld-schema-stack-complete-aeo-implementation-guide-2026" } ] } ```

2. SiteLinksSearchBox is a single schema block on the homepage that registers your site's internal search capability with search engines and AI crawlers. It has minimal direct AEO impact but contributes to entity completeness scoring. Implementation takes under ten minutes and should be done in week one alongside Organization schema.

Product and Service Schema: The Commerce Layer

For companies with product pages, SaaS pricing pages, or service offering pages, Product and Service schemas are the citation surfaces that answer "does product X do Y" and "what does service X cost" queries. These are high-value queries in the B2B buying cycle — exactly the queries where AI assistants are now influencing decisions before buyers visit the site.

```json { "@context": "https://schema.org", "@type": "SoftwareApplication", "name": "Signal Analytics Platform", "applicationCategory": "BusinessApplication", "operatingSystem": "Web", "description": "AI search citation tracking for B2B marketing teams. Measures share of model, citation accuracy, and competitor citation gaps across ChatGPT, Perplexity, Claude, and Gemini.", "offers": { "@type": "Offer", "priceCurrency": "USD", "price": "299", "priceValidUntil": "2026-12-31", "availability": "https://schema.org/InStock" }, "featureList": [ "Multi-engine citation tracking", "Share of model measurement", "Competitor citation gap analysis", "Citation accuracy auditing" ], "publisher": { "@type": "Organization", "name": "Signal", "url": "https://readsignal.io" } } ```

The featureList property is underused and high-value for AEO. AI models extract feature lists and use them to answer "does X support Y" queries directly. Comprehensive, accurate feature lists reduce the probability of AI-generated claims about your product being wrong — which matters both for citation quality and for prospect trust when the AI answer diverges from reality.

Keep pricing data current. AI models cache product information, and stale pricing data in schema creates trust damage when a prospect is quoted one price by an AI assistant and sees another on your site. If pricing changes frequently, use the priceValidUntil property to signal expiry and force AI model refresh.

For context on how AI agents are beginning to use product schema data for autonomous purchasing decisions, see agentic commerce and AI buying agents.

The Validation and Testing Workflow

Schema implementation without a validation workflow produces a false sense of security. Syntax-valid JSON-LD can still fail to generate citations for reasons that syntax validators don't catch. The complete validation workflow has four layers.

Layer 1: Syntax validation. Use Google's Rich Results Test (search.google.com/test/rich-results) for schema types that generate Google rich results (FAQPage, HowTo, Article, Product, BreadcrumbList). Use schema.org/validator for types that don't generate rich results. Fix all errors before proceeding.

Layer 2: Server-render confirmation. Run `curl -s https://yourdomain.com/page | grep -A 50 'application/ld+json'` to confirm your JSON-LD block appears in the raw HTML response. If it doesn't, your schema is client-rendered and invisible to AI crawlers. Fix SSR before anything else.

Layer 3: Field completeness audit. Check each schema block against the required and recommended fields in the schema.org documentation for that type. Common missed fields: dateModified on Article, sameAs on Organization and Person, acceptedAnswer on FAQPage Question objects, position on HowToStep. Missing recommended fields don't break schema but reduce citation confidence scores in AI models.

Layer 4: Live citation testing. This is the only validation that measures actual AEO output. Ask ChatGPT, Claude, and Perplexity with web browsing enabled to answer questions that your FAQPage schema addresses. Track whether your site is cited, whether the citation is accurate, and whether the answer text matches your acceptedAnswer text. Run this test monthly as a baseline.

Common Schema Bugs That Silently Kill AEO Performance

Several implementation errors consistently appear in AEO audits without triggering syntax validation errors:

Relative URLs in sameAs arrays. The sameAs property must contain absolute URLs. "/about" will not resolve correctly — it must be "https://yourdomain.com/about".

FAQPage nested inside WebPage. Some implementations nest FAQPage as a child of WebPage, which breaks extraction in several AI crawler implementations. FAQPage should be a top-level schema type or coexist at the same level as WebPage in a @graph array.

Missing mainEntity linking FAQPage to WebPage. Without the mainEntity relationship, FAQPage and WebPage schema blocks on the same page are treated as unrelated schemas. Use a @graph array with @id references to link them explicitly.

Duplicate schema blocks from CMS plugins. Yoast, RankMath, and other SEO plugins sometimes generate their own Article and FAQPage schemas that conflict with custom JSON-LD. AI crawlers encountering two conflicting Article schemas for the same page discount both. Audit your page source for duplicate schema and disable CMS-generated schema when adding custom implementations.

Schema on JavaScript-rendered pages. Covered above, but worth repeating: schema in SPAs that don't SSR is invisible. This is the single most common schema implementation failure across the sites we audit.

Implementing Schema at Scale: The Priority Order

The implementation priority order for teams shipping schema across a full site, not just individual pages:

1. Organization schema on homepage — single implementation, high impact, anchors all downstream entity graph. Estimated time: 2 hours including sameAs research.

2. Article schema on all content pages — typically implemented via CMS template, affects citation credibility sitewide. Estimated time: 4-8 hours with template implementation.

3. Person schema for all named authors — author profile pages get schema, referenced in Article publisher fields. Estimated time: 1-2 hours per author, worth building as a CMS field.

4. FAQPage schema on top 50 pages by traffic — requires content work (writing self-contained Q&A pairs) not just technical implementation. Estimated time: 2-3 hours per page including content writing.

5. BreadcrumbList sitewide — typically CMS-templated, low implementation time. Estimated time: 4 hours with template.

6. HowTo schema on procedural content — selective implementation, only pages with genuine step-by-step content. Estimated time: 1-2 hours per page.

7. Product/SoftwareApplication schema on product and pricing pages — requires accurate product data and maintenance commitment. Estimated time: 4-6 hours per product.

8. SiteLinksSearchBox on homepage — single implementation, low time, low impact but zero cost. Estimated time: 30 minutes.

The full implementation of this stack on a 500-page site takes approximately 6-8 weeks with one technical implementer and one content person handling FAQPage writing. The ROI starts materializing in citation rate improvements at approximately 4-6 weeks post-implementation, with full effect visible at 12-16 weeks as AI crawlers complete index refresh cycles.

For the measurement framework that lets you track citation rate changes as you implement each schema type, see the AEO citation tracking playbook and the broader discussion in share of model measurement without vanity metrics.

Schema Maintenance: The Discipline That Compounds

Schema implementation is not a one-time project. It is infrastructure that degrades without maintenance. The failure modes are different from the initial implementation failures — they are slow, invisible, and compounding.

Product schema pricing drift. Product and SoftwareApplication schemas with specific prices require updating every time pricing changes. AI models that cached the old price continue citing it until they recrawl. Set calendar reminders to audit product schema every time pricing changes.

dateModified staleness. Article schemas that are not updated when content changes signal staleness to AI models. Build dateModified updates into your content workflow — every substantial content edit should trigger a dateModified update.

Author sameAs link rot. Person schema sameAs URLs pointing to deleted or changed social profiles generate 404s that reduce entity verification confidence. Audit author Person schemas quarterly.

Organization sameAs coverage gaps. As your organization gains new web presences — a Wikipedia article, a Crunchbase profile, an industry directory listing — add them to your Organization sameAs array. Each new verified reference strengthens entity graph confidence.

Schema after platform migrations. CMS migrations, redesigns, and platform changes frequently break schema implementations that weren't built into the migration checklist. Schema validation should be an explicit deliverable in any platform migration project, not an afterthought.

The teams running the best AEO schema programs in 2026 treat schema maintenance the way they treat link health — as an ongoing operational function, not a shipped feature. They run automated validation checks weekly, flag drift in citation accuracy as an indicator of schema degradation, and have clear ownership of schema updates when content or product details change.

For a full treatment of how schema fits into the broader technical AEO infrastructure — including AI crawler rendering, robots.txt configuration, and crawl budget allocation — see Google AI Overviews and the AEO mandate for publishers.

The Schema Stack Checklist

A summary of what complete schema implementation looks like, in copy-paste checklist format:

1. Homepage — Organization (with sameAs), WebSite, SiteLinksSearchBox

2. Every article/editorial page — AnalysisNewsArticle or NewsArticle, Person (author), BreadcrumbList, FAQPage (5+ Q&A pairs)

3. Every product/service page — SoftwareApplication or Service, Offer (with current pricing), BreadcrumbList

4. Every author profile page — Person (with knowsAbout, worksFor, sameAs)

5. Every how-to/tutorial page — HowTo (with ordered HowToStep), Article, FAQPage

6. Every category/tag page — BreadcrumbList, CollectionPage

7. All pages — confirm JSON-LD appears in server-rendered HTML via curl test before deploying

The schema stack described here is not aspirational. The sites running it today — established trade publications, well-structured SaaS documentation, organized B2B content hubs — are measurably outperforming sites with partial implementations in AI citation rates across ChatGPT, Perplexity, and Claude. The implementation window before AI citation defaults harden is measured in quarters, not years. Every month of schema incompleteness is a month of citations accruing to competitors who shipped the stack first.

Takeaway: Most AEO schema programs are 20% complete — Article and FAQPage with nothing underneath. The other 80% is the entity graph layer: Organization schema with verified sameAs links, Person schema for author authority, BreadcrumbList for structural context, HowTo for procedural content, and Product schema for commercial pages. Complete implementation takes 6-8 weeks on a typical site and produces measurable citation rate improvement at 4-6 weeks post-launch. The maintenance discipline — keeping prices current, updating dateModified, auditing sameAs links — is what separates programs that compound their advantage from programs that degrade after the initial lift. Ship the stack, build the maintenance workflow, and measure citation rate monthly. That is the entire playbook.