FAQPage Schema the Right Way: JSON-LD Only

The Three Rules for FAQPage Schema

The most common FAQPage schema failure on WordPress is duplicate structured data output. RankMath, Yoast, and other SEO plugins auto-generate FAQPage JSON-LD from post content, producing a second FAQPage declaration alongside any manually added schema. On a WordPress site with RankMath or any schema plugin active, there may be two FAQPage declarations in the rendered output — one from your custom JSON-LD, one from the plugin. Both are valid individually. Together they trigger SCHEMA_FAQPAGE_DUPLICATE_001. GSC emails you about it 3–5 days later.

The Correct JSON-LD Pattern

Full FAQPage JSON-LD structure as implemented across thegeolab.net — five question-answer pairs, each answer written to stand alone as a complete response to the query:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "What is generative engine optimisation?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Generative engine optimisation (GEO) is the practice of designing
content to be retrieved, extracted, and cited by AI search systems.
Unlike traditional SEO, which optimises entire pages for position in
ranked results, GEO optimises individual content sections for inclusion
in AI-generated answers."
      }
    },
    {
      "@type": "Question",
      "name": "How is GEO different from SEO?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "SEO optimises for rankings and click-through rate using page-level
signals like backlinks and domain authority. GEO optimises for retrieval
and citation rate using section-level signals like extractability and
entity clarity. The optimisation unit is different: SEO is
document-centric, GEO is retrieval-centric."
      }
    }
  ]
}
</script>

And the pattern to avoid — microdata attributes on the FAQ HTML that, combined with the JSON-LD above, produce a duplicate declaration:

<!-- This creates a second FAQPage declaration -->
<div itemscope itemtype="https://schema.org/FAQPage">
  <div itemscope itemprop="mainEntity" itemtype="https://schema.org/Question">
    <h3 itemprop="name">What is GEO?</h3>
    <div itemscope itemprop="acceptedAnswer" itemtype="https://schema.org/Answer">
      <p itemprop="text">GEO is...</p>
    </div>
  </div>
</div>

Writing FAQ Questions and Answers

Schema structure is only half the implementation. The questions and answers themselves determine whether the schema is useful as a retrieval signal.

Questions

Write questions the way a user would phrase a query to ChatGPT, Gemini, or Perplexity — not the way you’d write a heading. “What is FAQPage schema?” is a heading. “Does FAQPage schema still work in 2026?” is a query. The question in the schema should match the retrieval query you want to rank for.

Answers

Per the FAQPage schema documentation, each answer should stand alone as a complete, accurate response to the question — without requiring the reader to have read anything else on the page. AI retrieval systems extract individual Q&A pairs. If the answer references context from earlier in the page (“as mentioned above”), that context won’t be present when the pair is extracted in isolation.

The working range for thegeolab.net is 5–10 pairs per page. Below 5, the schema adds minimal retrieval signal. Above 10, individual answers tend to shorten and become less substantive. The target is answers long enough to be genuinely useful — typically 2–4 sentences — not answers optimised for brevity.

Implementation on WordPress

On a WordPress site, FAQPage JSON-LD can be output in three ways. Only one of them is safe to use alongside a schema plugin like RankMath:

1. Via functions.php using wp_head Add the JSON-LD via add_action('wp_head', 'output_faqpage_schema'). Gives you full control over which pages receive the schema and what it contains. Safe alongside RankMath as long as RankMath’s own FAQ schema output is disabled.
2. Inline in post content as a <script> block Placed directly in the post content, inside a <!-- wp:html --> wrapper to prevent wpautop from processing it. Works for posts where the FAQ is page-specific. Risk: if functions.php also outputs FAQPage schema for that page, you’ll have a duplicate.
3. Via RankMath’s FAQ block RankMath outputs FAQPage JSON-LD when you use its FAQ block in the editor. Safe only if you have no other FAQPage schema on the page. If you also have custom JSON-LD from functions.php or inline in the content, disable RankMath’s FAQ schema output entirely.

On thegeolab.net: FAQPage JSON-LD is output via functions.php for core pages and the homepage, and inline in v3 post content for individual posts. RankMath’s FAQ block is not used. audit.py Module 6 verifies FAQPage schema coverage across functions.php (checking for a minimum of two instances) and flags any page with a duplicate declaration.

How to Verify Your Implementation

Three checks, in order of speed:

The GEO compliance audit system includes a dedicated FAQPage verification check — the 71-check audit system flags duplicate schema output, missing answer length, and JSON-LD graph conflicts that the Google Rich Results Test does not catch.

curl -s https://yourdomain.com/your-page/ | grep -c "FAQPage"

Output should be 1. Output of 2 or higher means a duplicate declaration is present in the rendered HTML.

https://search.google.com/test/rich-results

Enter your page URL. The tool will show all detected schema declarations. Two FAQPage entries means a duplicate. One entry, correctly structured, means the implementation is clean.

Google Search Console → Enhancements → FAQ. Any “Duplicate field” errors will appear here — but this check is the slowest, as GSC may take days to process newly deployed pages.

The curl | grep check takes under 10 seconds and should be run after every deployment that touches schema. It’s the last step in the v3 deployment checklist for a reason.

Frequently Asked Questions

What Practitioners Are Saying