TL;DR

Google deprecated HowTo rich results in 2023 and FAQPage rich results on May 7, 2026. But the underlying schema types still exist and still aid machine comprehension. Five types deliver the clearest ROI in 2026: Article/BlogPosting, Product, Organization, LocalBusiness, and VideoObject. Everything else is situational. Implement in JSON-LD. Audit and fix errors quarterly.

By Guru Editorial | June 15, 2026

Schema markup promised a simple trade: add machine-readable metadata, get enhanced search listings in return. For years that trade held up. In 2026 it has become more complicated, and more interesting.

Google AI Overviews now cover a large share of informational queries, and Ahrefs data shows they reduce position-1 organic click-through rates by 58%, with zero-click searches jumping from 54% to 72% of all queries. At the same time, Google has stripped two of the most heavily-used rich-result types off the SERP entirely. If you built your structured-data strategy around FAQ or HowTo snippets, you are working off a plan that is two rewrites out of date.

This guide covers what schema markup actually does in 2026, which types still produce a measurable SERP or AI-visibility benefit, which are deprecated or oversold, and how to build a defensible implementation that holds up through the next round of changes.

What Schema Markup Does (and Does Not Do) in 2026

Schema markup is a standardized vocabulary, maintained at Schema.org, that lets you annotate page content with machine-readable labels. You wrap a product name, a review rating, or a business address in structured markup so that search engines, AI crawlers, and knowledge-graph systems can extract facts without parsing natural language.

What schema markup does not do is rank your page higher. Google has stated this clearly and repeatedly: structured data is an eligibility signal, not a ranking signal. Adding Article schema to a thin post will not push it up the results. What it can do is make a well-qualified page eligible for visual enhancements, Knowledge Panel data, or rich-result features that improve click-through rate once the page is already ranking.

The second misconception worth retiring: schema does not directly influence most AI citations. A May 2026 Ahrefs study tracked 1,885 pages that added JSON-LD between August 2025 and March 2026, matched against 4,000 control pages, and found no statistically significant increase in citations from Google AI Overviews, Google AI Mode, or ChatGPT. Pages with schema are more frequently cited, but Ahrefs interprets that as a quality-of-site correlation, not a causal schema effect. The Otterly AI experiment reached a similar conclusion: when AI systems retrieve a page in real time, they read visible HTML, not the JSON-LD block.

This does not mean schema is useless for AI. It means the relationship is indirect. Structured data improves Knowledge Graph entity recognition, feeds Bing and third-party crawlers that have their own citation behaviors, and, as we cover below, still drives SERP rich results that generate clicks. Those clicks matter more than ever when AI Overviews are cannibalizing zero-click impressions.

The Deprecation Timeline You Need to Know

Understanding what has been removed is as important as knowing what still works.

HowTo rich results: removed August 2023. Google stopped showing the step-by-step HowTo display in standard search results. The schema type remains valid, and you can keep the markup, but it no longer earns an enhanced SERP appearance on Google. Other crawlers, including Bing, continue to use it.

FAQPage rich results: removed May 7, 2026. Google confirmed the removal through the Google Search Status Dashboard. The FAQ accordion that previously appeared beneath certain SERP listings is gone. As with HowTo, the FAQPage schema type itself remains a valid Schema.org vocabulary. Google's own documentation says it will continue using FAQPage markup to understand page content. However, you should not expect any visual SERP enhancement from it on Google going forward.

Speakable schema: removed from Google's rich results support. Never gained significant adoption and Google no longer lists it in its supported types documentation.

For a full picture of what survives and what does not, the comparison table below maps each major schema type to its current 2026 status.

Schema Type Comparison: What Still Earns Rich Results

Table: Current 2026 status for the most-deployed schema types.

Schema TypeGoogle Rich ResultAI/Knowledge Graph ValuePriority
Article / BlogPostingTop Stories carousel, DiscoverHigh (entity grounding)High
Product + OfferPrice/availability snippet, ShoppingHigh (e-commerce extraction)High
AggregateRatingStar ratings in SERPMediumHigh (where eligible)
OrganizationKnowledge Panel, logo in SERPsHigh (brand entity)High
LocalBusinessMap Pack data, local Knowledge PanelHighHigh (local)
VideoObjectVideo carousel, Key MomentsHighHigh (video content)
BreadcrumbListBreadcrumb display in SERP URLMediumMedium (all sites)
EventEvent rich results, Google Events tabHighMedium (event pages)
RecipeRecipe rich results, image carouselsHighHigh (food sites)
FAQPageNone (deprecated May 2026)Low-mediumLow (keep, do not add)
HowToNone (deprecated Aug 2023)LowLow (keep, do not add)
JobPostingJob rich resultsHighHigh (job listings)
SoftwareApplicationApp rating snippetMediumMedium (SaaS)

The table reflects what Google's own Rich Results Test and documentation support as of June 2026. "AI/Knowledge Graph Value" reflects utility for entity disambiguation and third-party crawler comprehension, independent of Google's SERP rich results.

The Five Schema Types That Pay Off in 2026

1. Article and BlogPosting

Article schema is the foundation for content-focused sites. Implementing it correctly, with author, datePublished, dateModified, headline, and image populated, makes your editorial content eligible for the Top Stories carousel and Google Discover. Both placements can drive significant incremental traffic that organic ranking position alone does not capture.

The author field has become more important since Google's documentation on E-E-A-T guidance expanded. A named author linked to a Person entity with a sameAs reference to their Google Scholar, LinkedIn, or About page creates a verifiable signal that AI engines use when assessing source credibility. For more on building those signals, see our guide to building E-E-A-T signals that Google and AI engines actually trust.

A populated dateModified field is also underused. It signals content freshness to Googlebot and to AI crawlers performing real-time retrieval. If you refresh an article, update dateModified. This is a one-field change with clear upside.

2. Product Schema with Offer and AggregateRating

For e-commerce, Product schema is non-negotiable. It determines whether your product pages display price, availability, and star ratings directly in Google Shopping results and in the standard organic SERP. Pages with Review and AggregateRating markup consistently show meaningfully higher click-through rates than identical listings without those signals, with industry analyses frequently citing gains in the 20-35% range.

The critical fields are name, description, image, sku, offers (with price, priceCurrency, and availability), and aggregateRating (with ratingValue and reviewCount). Missing any of the required Offer fields typically disqualifies the page from the price-snippet display. Review count matters: Google generally requires more than a handful of reviews before displaying AggregateRating in results.

One common mistake is using Organization or WebSite schema on product pages instead of Product schema, or nesting AggregateRating outside the Product entity. Run every product page type through Google's Rich Results Test before deploying at scale.

3. Organization Schema

Organization schema belongs on your homepage and, ideally, your About page. It defines your brand entity: legal name, logo, social profiles, contact point, and URL. This data populates your Knowledge Panel and is the primary signal that entity disambiguation systems use to link mentions of your brand across the web.

The sameAs property is the workhorse here. Linking your Organization entity to your Wikidata entry, Crunchbase profile, LinkedIn company page, and other authoritative directories tells every knowledge graph system, including those powering AI Overviews and ChatGPT's browsing model, that these references all point to the same real-world entity. For SaaS and agency sites, this is the structured-data change with the highest Knowledge Graph return per hour of implementation time.

4. LocalBusiness Schema

For local pages, LocalBusiness schema (or one of its more specific subtypes like LegalService, MedicalBusiness, or Restaurant) feeds the data that populates Google Maps, the local Knowledge Panel, and near-me AI answers. The required fields are name, address (using PostalAddress with full street, city, state, and zip), telephone, url, and openingHoursSpecification.

Consistency between your LocalBusiness schema data and your Google Business Profile is essential. Mismatches in address format, phone number, or business name signal contradictory information and reduce the reliability score Google assigns to your entity data. This is one of the most common errors we find when running technical SEO audits for multi-location clients.

5. VideoObject Schema

Video-heavy pages are one of the clearest schema wins still available in 2026. VideoObject markup makes your video eligible for the Video rich result carousel, the dedicated Video search tab, and, if you populate the hasPart Clip property with timestamps, the Key Moments feature. Key Moments display your video chapters directly on the SERP and allow searchers to jump to a specific point, which measurably increases both impressions and engagement.

Required fields: name, thumbnailUrl, and uploadDate. Recommended: description, duration, contentUrl or embedUrl, and hasPart with Clip objects for each chapter. A Google Search Central case study on Vidio, the Indonesian streaming platform, found that implementing VideoObject markup delivered approximately 3x more video impressions and close to 2x more video clicks within a year, even accounting for a roughly 30% increase in published video volume over the same period.

The Nuanced Cases: FAQPage, HowTo, and BreadcrumbList

Schema Type Decision Framework: 2026 Schema Type SERP Rich Result AI/KG Utility Action Article / BlogPosting Top Stories, Discover High Implement Product + AggregateRating Price/Stars snippet High Implement FAQPage None (deprecated May 2026) Low-Medium Keep; Don't Add New HowTo None (deprecated Aug 2023) Low Keep; Don't Add New Organization / LocalBusiness Knowledge Panel High Implement

Schema type decision matrix for 2026: what to implement, what to keep, and what to deprioritize.

FAQPage: Do not remove existing FAQPage markup from pages where the content genuinely answers questions. Google has confirmed it will continue using the markup to understand pages, and Bing and RAG crawlers have their own behaviors. But do not invest time adding FAQPage markup to new pages with the expectation of a rich result, because that return is gone on Google.

HowTo: Same guidance as FAQPage. Existing markup stays; new implementations are not worth prioritizing for Google rich results. If you are building structured how-to content and want AI comprehension, put more effort into clear visible headings and numbered lists in the HTML body, which every AI retrieval system reads directly.

BreadcrumbList: Still supported, still displays breadcrumb navigation in the SERP URL line instead of a plain URL. This is a quick win for most sites, costs almost nothing to implement, and makes SERP listings marginally more readable. Implement it site-wide via your CMS or site template.

Schema and GEO: The Indirect Connection

Generative engine optimization (GEO) is the practice of making content more legible and citable to AI answer engines, a domain we cover in depth in our GEO-focused tools and scoring. The question SEOs are asking in 2026 is whether schema helps with GEO specifically.

The honest answer, supported by the Ahrefs May 2026 study, is that adding JSON-LD schema to an otherwise unchanged page produces no measurable increase in AI citations. But that framing misses the more important point.

Schema is one layer of a content-quality stack. Pages that AI engines cite consistently share several properties: clear entity markup, visible authorship, cited statistics, structured prose with explicit headings, and domain authority built through external links. The Princeton GEO study (KDD 2024) found that adding statistics increased GEO scores by 41%, adding citations increased them by up to 115% for low-ranking pages, and adding quotations increased them by 28%. Schema supports those signals by making entity relationships machine-legible, but it does not substitute for the underlying content quality.

Put differently: schema is a prerequisite for Knowledge Graph entity recognition, and entity recognition does influence which sources AI systems default to trusting. Organization schema with strong sameAs references is the most direct schema-to-GEO connection available. For a full treatment of optimizing for both Google and AI engines simultaneously, see our guide to SEO + GEO optimization.

A Practical Implementation Checklist

Schema Implementation Priority Order Step 1 Organization Homepage + About Step 2 BreadcrumbList All site pages Step 3 Article/BlogPosting All editorial content Step 4 Page-type specific Product/Local/Video Page-type specific schema (Step 4 detail): E-Commerce Product + Offer + AggregateRating Local Business LocalBusiness + ContactPoint Video Content VideoObject + hasPart (Key Moments) Step 5: Quarterly Validation Rich Results Test + GSC Enhancements report + Search Appearance audit

Five-step schema implementation sequence: build the foundation first, then add page-type specific types, then validate quarterly.

The following checklist covers everything needed for a complete 2026 schema implementation. Use it as a deployment gate before any new page type goes live, and as an audit template for existing pages.

Foundation (all sites):

  • Organization schema on homepage with name, url, logo, and sameAs pointing to at least three authoritative external profiles
  • BreadcrumbList on all interior pages, reflecting actual URL hierarchy
  • WebSite schema with SearchAction if your site has a search function
  • JSON-LD format in the <head> for all markup (not inline Microdata)

Content pages:

  • Article or BlogPosting on all editorial URLs with author (Person entity with sameAs), datePublished, dateModified, headline, and image
  • Keep FAQPage markup where it already exists; do not invest in new FAQPage implementations for Google rich results

E-commerce pages:

  • Product schema with complete Offer (price, currency, availability, URL) and AggregateRating (ratingValue, reviewCount)
  • Verify in Rich Results Test before deploying product template changes

Local pages:

  • LocalBusiness (or specific subtype) with full PostalAddress, telephone, and openingHoursSpecification
  • Cross-check every field against your Google Business Profile for consistency

Video pages:

  • VideoObject with name, thumbnailUrl, uploadDate, duration, and hasPart Clip objects for each chapter

Validation and maintenance:

  • Run Google's Rich Results Test on one representative URL per page type monthly
  • Check Google Search Console Enhancements report for errors and warnings quarterly
  • Update dateModified on Article/BlogPosting every time content is substantively revised
  • Re-audit any page type after a CMS template change, since template changes frequently break JSON-LD generation

The Guru technical SEO tools include automated schema validation that surfaces errors at the page level and flags when a template change breaks structured data across a page group, so you catch regressions before they compound.

Common Schema Mistakes That Quietly Kill Your Rich Results

Even sites that implement schema thoughtfully tend to make a handful of recurring mistakes that either prevent rich results or trigger manual actions.

Marking up content that is not visible on the page. Google's structured data guidelines are explicit: do not use schema to describe content that users cannot see. If your star rating schema contains a ratingValue that does not correspond to visible review text, that is a policy violation, not a gray area.

Stale datePublished and dateModified values. If a page was published in 2022 and has a datePublished of 2022 but dateModified has never been updated despite multiple content revisions, it signals to both Google and AI crawlers that the content has not changed. Update dateModified whenever you make substantive edits.

Using AggregateRating with fewer than five reviews. Google typically requires a minimum review count before rendering star ratings. The threshold is not officially published, but five is a practical floor based on observed behavior.

Mixing schema types across multiple JSON-LD blocks incorrectly. Placing your Organization markup in one script tag and your Article markup in another is fine. But nesting unrelated entities incorrectly (for example, putting an AggregateRating inside an Article instead of a Product) produces errors that silently disqualify the page from rich results.

Not testing after CMS updates. This is the most common source of schema regressions at scale. A template change rolls out, the JSON-LD generation logic breaks or drops a required field, and hundreds of product pages quietly lose their price snippet. Integrate Rich Results Test validation into your deployment checklist. Our guide on on-page SEO factors that still move rankings in 2026 covers where to slot this into a broader technical review process.

What to Prioritize by Site Type

The right schema priority stack depends on your site type. These recommendations reflect what delivers the highest expected return given the 2026 SERP and AI landscape.

SaaS and software sites: Organization (with sameAs), Article/BlogPosting for the blog, SoftwareApplication for product pages if applicable. Focus on entity establishment; rich results are limited for most SaaS content types.

E-commerce: Product + Offer + AggregateRating is the single highest-ROI schema investment available. BreadcrumbList and Organization round out the foundation. The CTR lift from star ratings and price snippets is direct and measurable.

Local businesses and agencies: LocalBusiness with complete address and hours is the foundation. Organization for brand entity, Article for any blog or resource content.

Content and media sites: Article/BlogPosting, VideoObject if video is published, and Event if events are covered. These are the types that unlock Top Stories, Discover placement, and the Video carousel.

Law and medical: Person schema for attorney and physician profiles, with professional credential markup, pairs with LocalBusiness and Article. The Guru tools for law firms and medical practices include templates for these specific schema patterns.

For a full breakdown of which SERP features are realistically achievable by page type and how to target them, see our guide on winning featured snippets and SERP features in 2026.

Frequently Asked Questions

Does schema markup help with Google rankings?

No. Google has confirmed repeatedly that structured data is not a ranking factor. It is an eligibility signal: it makes pages eligible for visual enhancements and rich results, which can improve click-through rate, but it does not change where a page ranks for a given query. The underlying content, links, and technical quality determine ranking.

Should I remove FAQPage schema now that Google deprecated the rich result?

No. Google's official guidance says you can keep or remove it; either is fine. Other search engines and AI crawlers continue to process FAQPage markup. Removing it saves a few lines of code but produces no measurable improvement. Keep existing markup; do not add new FAQPage implementations expecting Google rich results.

Does schema markup directly increase AI citations from ChatGPT or Perplexity?

The evidence says no direct causal effect. A May 2026 Ahrefs study tracking nearly 1,900 pages found no statistically significant increase in AI citations after adding JSON-LD. Pages with schema are more frequently cited, but this reflects site quality, not schema itself. Schema remains valuable for entity recognition and Knowledge Graph presence.

What format should I use for schema markup: JSON-LD, Microdata, or RDFa?

JSON-LD, placed in the <head>. Google recommends it explicitly because it is the easiest to maintain, does not require modifying HTML body elements, and works cleanly with modern JavaScript-rendered sites. Microdata and RDFa are supported but not recommended for new implementations.

How do I validate that my schema is working correctly?

Use Google's Rich Results Test (search.google.com/test/rich-results) for individual URLs and Google Search Console's Enhancements section for site-wide error monitoring. The Enhancements report shows how many pages are eligible for each rich result type, how many have errors, and what those errors are. Check it quarterly at minimum.

Is HowTo schema worth implementing in 2026?

Not for the Google rich result, which was removed in August 2023. Keep existing HowTo markup since it costs nothing to retain and other crawlers still use it. For new instructional content, write clearly structured numbered steps in visible HTML, which AI retrieval systems read directly.

What is the single highest-impact schema change for a site that has done nothing yet?

Organization schema on the homepage, fully populated with sameAs links to authoritative external profiles. It takes under an hour to implement, establishes your brand as a recognized entity in Google's Knowledge Graph, and provides the entity foundation that every other schema type builds on. Do this before anything else.

How often should I audit schema for errors?

At minimum, quarterly, using the Google Search Console Enhancements report. Additionally, run the Rich Results Test on your core page templates any time a CMS, theme, or plugin update is deployed, since template changes are the most common cause of schema regressions at scale.

Sources