since Google FAQ rich results were officially deprecated in May 2026, with many SEOs concluding that FAQPage markup no longer mattered.

That interpretation misses the bigger picture.

FAQ rich snippets There was only one application layer built on top of the broader semantic infrastructure. What was missing was a visual SERP feature. What remains is the semantic layer beneath it.

This distinction matters even more now as search evolves toward AI systems, retrieval pipelines, and machine-readable publishing.

The Web is becoming increasingly agent-consumable, and structured meaning is more important than ever.

A brief history of FAQ page markup

When Google introduced FAQ rich results, this idea made perfect sense.

Web pages already include question-answer sections. FAQ page markup allowed publishers to clearly structure those relationships so that search engines could:

  • understand the material better,
  • Find out the short answer,
  • Improve SERP Presentation,
  • And potentially helps users solve queries faster.

At first, the results were impressive.

Pages in search results expanded dramatically. CTR improved. SEO plugins added one-click FAQ generator. Entire industries adopted FAQ markup almost overnight.

But something else also happened.

FAQPage quickly became one of the most commonly used forms of structured data on the Web.

Publishers started adding bloated FAQ sections to almost every imaginable page:

  • product page,
  • category page,
  • blog post,
  • landing page,
  • associated Content,
  • Thin SEO pages.

Many FAQs were not written for users. They were written for SERP expansion.

The pattern became familiar:
The useful Google feature is used extensively until the signal quality deteriorates.

“Misuse it, lose it” warning

Long before the final shutdown, lily ray And other experienced SEOs warn that FAQ rich results were heading towards abuse saturation.

The unofficial rules of modern search features became:

Abuse it, lose it.

Importantly, this was never an argument against Schema.org or structured data.

This was an argument against manipulative deployment patterns.

Google's problem wasn't semantic markup.

Google's problem was low-quality implementation at Internet scale.

This distinction is necessary because much of today's discourse still conflates the two.

Google's progress towards FAQ rich results has slowed down

Removing the FAQ snippet didn't happen overnight.

Google gradually reduced its visibility over several years.

First came the tightening of eligibility.

FAQ rich results then became more and more limited to official domains, specifically:

  • government sites,
  • health websites,
  • Highly trusted publisher.

Over time, many SEOs noticed that FAQ snippets became less visible even when the markup was technically valid.

Eventually, Google announced the final shutdown.

The important thing to understand is this:

Google minimized the rendering layer, not the Schema.org terminology.

The visual feature disappeared.

The semantic model did not do this.

May 7, 2025: End of FAQ Rich Results

On May 7, 2025, Google officially removed FAQ rich results from Search.

For many people, it became shorthand for:

“The FAQ schema is dead.”

But that statement confuses presentation with meaning.

Schema.org is not a SERP feature.

Schema.org is a shared vocabulary for expressing machine-readable meaning.

FAQ rich results were simply a user interface built on top of that semantic layer.

Removing the UI does not invalidate the semantics.

FAQWhy does page markup still matter?

The disappearance of FAQ rich snippets does not mean that the full value of FAQPage markup is lost.

In many ways the opposite is happening.

The Web is transitioning from a human-first retrieval layer to a hybrid ecosystem where machines increasingly consume, interpret, summarize, and recombine content.

This completely changes the role of structured data.

FAQ Page Markup for Traditional SEO

Even without visible rich snippets, FAQPage markup still helps search engines:

  • Understand clear question-answer relationships,
  • clarify topics,
  • structure information continuously,
  • strengthen unit associations,
  • and extract machine-readable meaning.

This does not mean that FAQPage markup automatically improves rankings.

There is no credible evidence of a direct increase in ranking.

But semantic clarity still matters.

Structured quiz content creates clean information structure and strong thematic organization.

Search engines continue to rely heavily on internally structured extraction pipelines, even when no visible SERP enhancements are shown.

The disappearance of a rich result does not mean the disappearance of machine understanding.

FAQ page markup for GEO and AI search

This is where the conversation becomes even more interesting.

Large language models and AI retrieval systems prioritize content that is:

  • clear,
  • Compressed,
  • structured,
  • machine-readable,
  • And it is easy to divide into retrievable units.

FAQ structures naturally meet those needs.

Question-answer formatting naturally aligns with:

  • recovery pipeline,
  • find the answer,
  • semantic chunking,
  • and interactive interface.

In that sense, FAQPage markup has moved from being primarily a SERP enhancement mechanism to being part of an AI-readable publishing layer.

This is a big change.

Over the years, many SEOs have optimized FAQPage for visual expansion in Google search.

Now the occasion is different:
Adaptation for machine understanding.

FAQSection by offer jost de valk

One of the limitations of the basic FAQPage model is that it assumes that the entire page is basically a FAQ.

But modern materials rarely work that way.

Most high-quality documents include:

  • story section,
  • editorial section,
  • product information,
  • reference,
  • And embedded FAQs.

This is part of the motivation behind the discussions around More granular semantic structures like FAQSection.

This idea is important because it reflects the broader developments taking place on the Web:
Moving from page-level semantics to a composable semantics section.

The future of semantic publishing probably doesn't look like this:

Instead, it is:

  • A document that contains several meaningful components.

This is very close to how humans and AI systems actually consume information.

From rich results to knowledge graphs: my own usage patterns

Personally, I'm using the FAQPage markup not as a separate SEO widget, but as part of a broader entity-centric semantic graph.

The most interesting patterns emerge when connecting:

  • FAQPage.about
  • through entity relationships Thing
  • and use reverse references subjectOf.

In this model, FAQs stop being standalone snippets and become meaningful statements about entities.

For example:

  • may be a product about FAQPage's goal,
  • While the product itself references the FAQPage subjectOf.

This creates a clear bidirectional semantic relationship.

The result is not just “markup”.

This becomes graph enrichment.

This is especially important in the AI-native web where systems increasingly rely on:

  • unit grounding,
  • relevant relationships,
  • semantic disjunction,
  • and retrieval-aware content structures.

The future of structured data is no different from rich results.

This is the associated meaning.

Schema.org as a memory layer for AI systems

One of the biggest misconceptions about Schema.org is that it is limited to “structured data for Google rich results.”

That was always the narrowest possible interpretation.

What is emerging now is huge:
Schema.org is rapidly becoming part of the memory layer for AI systems, agents, retrieval pipelines, and machine-readable applications.

The value is not limited to publishing markup on a web page.

The real value lies in creating interoperable semantic structures:

  • connect entities,
  • organize knowledge,
  • Enrich recovery,
  • And provide machine-readable grounding for AI systems.

This becomes especially powerful when semantic relations are combined with vector search and retrieval systems.

For example, one of the patterns I use is a combination of:

  • Schema.org entity relationships,
  • semantic chunking,
  • vector similarity search,
  • and GraphQL retrieval.

Instead of retrieving isolated text embeddings, the system can retrieve semantically linked information:

  • Answer,
  • their parents question,
  • FAQ page,
  • And the entity the page is actually about.

This changes retrieval from “same text lookup” to contextual semantic navigation.

Here's a simplified example of that pattern:

{
  entitySearch(
    query: {
      typeConstraint: { in: ["http://schema.org/Answer"] }
      search: { chunkset: "answer-1", string: "What is translated" }
    }
  ) {
    iri
    score: float(name: "_:score")
    entity: resource(name: "seovoc:isChunkOf") {
      type: string(name: "rdf:type")
      text: string(name: "schema:text")
      question: resource(name: "schema:parentItem") {
        iri
        type: string(name: "rdf:type")
        name: string(name: "schema:name")
        faq_page: resource(name: "schema:parentItem") {
          iri
          type: string(name: "rdf:type")
          name: string(name: "schema:name")
          text: string(name: "schema:text")
          about: resource(name: "schema:about") {
            iri
            type: string(name: "rdf:type")
            name: string(name: "schema:name")
            markdown_text: string(name: "seovoc:markdownText")
            url: string(name: "schema:url")
          }
        }
      }
    }
  }
}

What matters is that semantic relationships become traversable memory structures.

Embedding helps in retrieving relevant parts.

The graph provides context, grounding, interpretability, and semantic continuity.

This is where structured data stops being “SEO markup” and starts becoming the infrastructure for AI-native applications.

final thoughts

FAQ Rich snippets disappeared as the industry aggressively optimized for visual SERP expansion.

But the broader trajectory of the Web points in the opposite direction:
Towards more machine-readable meaning, not less.

As AI systems become increasingly involved in retrieval, synthesis, and answer generation, semantic publishing becomes the fundamental infrastructure.

This means that the actual value of the FAQPage markup was probably misunderstood from the beginning.

The snippet was temporary.

Semantics are long term.

And in an AI-native web, shared vocabularies like Schema.org may be more important than ever.