← Back to blog

How to Add Schema Markup for AI Citations (with FAQ Schema Generator Walkthrough)

FactSentry Team

5/2/2026

#schema#structured-data#geo

Schema markup is one of the highest-ROI things you can ship for AI search visibility. Five minutes per page of JSON-LD raises citation rate measurably, and the underlying schemas haven't changed in years. Despite that, most SaaS sites still ship with no structured data, partial schema, or a wrong implementation that does nothing.

This post walks through how to add schema markup for the four schemas that actually matter to AI engines: Article, FAQPage, Product, and Organization. With a FAQ schema generator pattern at the end you can copy.

Schema is one piece of the answer engine optimization playbook; for the writing-side rules see AI content optimization and LLM SEO.

Why schema matters for AI citations

AI engines read body copy, but they trust structured data. Article schema tells the model "this page is an article about X by author Y." FAQPage schema tells it "these are explicit Q&A pairs you can quote verbatim." Product schema tells it "this is a product, here's the price, here's what it does."

The pattern: where body copy says "we believe," structured data says "this is a fact." The model treats the second more confidently and cites it more often.

A second reason: structured data persists across redesigns. Your hero copy changes every six months. Your Organization JSON-LD doesn't. The model learns the structured data and treats it as ground truth.

The four schemas every SaaS site should ship

In rank order of impact:

  1. Organization — once, on every page. Tells the model who you are.
  2. Product — on the homepage and pricing page. Tells the model what you sell and what it costs.
  3. FAQPage — anywhere you have a Q&A block. Highest single-page citation lift.
  4. Article — on every blog post. Gives the model author, date, and topic.

Three more are worth shipping in specific cases:

  • BreadcrumbList — on deep pages.
  • HowTo — on tutorial-style content.
  • Review / AggregateRating — only if you have real reviews to cite.

How to add schema markup: the working pattern

Schema goes in JSON-LD form inside a <script type="application/ld+json"> tag in your page's <head> or anywhere in <body>. AI engines parse both.

For Next.js (App Router), the cleanest pattern is to render the script in your page component:

const jsonLd = {
  "@context": "https://schema.org",
  "@type": "Article",
  headline: "Your post title",
  author: { "@type": "Person", name: "Your Name" },
  datePublished: "2026-05-02",
};

return (
  <>
    <script
      type="application/ld+json"
      dangerouslySetInnerHTML={{ __html: JSON.stringify(jsonLd) }}
    />
    <article>{/* your content */}</article>
  </>
);

For static HTML, paste the same <script> tag in the <head>.

For WordPress, RankMath and Yoast handle most of this. You'll usually still want to hand-write the FAQPage schema for any custom Q&A blocks.

FAQ schema generator pattern

Here's the FAQPage JSON-LD pattern. Copy it, replace the questions and answers, ship.

{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "What is the question your buyer types?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Your literal answer in one to three sentences. This is the text the model will quote, so write it the way you'd want quoted."
      }
    },
    {
      "@type": "Question",
      "name": "Second question",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Second answer."
      }
    }
  ]
}

That's the entire FAQ schema generator. There's no faster way; the JSON is the JSON. If you want a UI tool to generate it, the schema.app validator and Google's Rich Results Test both render this form correctly.

Five rules for FAQPage schema that get cited more often:

  1. Use the literal questions your buyer types into ChatGPT, not internal jargon.
  2. Keep each answer 1–3 sentences. Long answers don't get quoted; they get summarised.
  3. Lead the answer with the answer itself — no "great question."
  4. Match the schema to a visible FAQ block on the page. Hidden schema is a Google penalty risk and AI engines downweight it too.
  5. Update dates when answers change. The model trusts fresher content.

Product schema for SaaS

Most SaaS sites get Product schema slightly wrong. Here's a clean version for a SaaS product with multiple price tiers:

{
  "@context": "https://schema.org",
  "@type": "SoftwareApplication",
  "name": "Acme",
  "applicationCategory": "BusinessApplication",
  "operatingSystem": "Web",
  "offers": [
    {
      "@type": "Offer",
      "name": "Starter",
      "price": "19",
      "priceCurrency": "USD"
    },
    {
      "@type": "Offer",
      "name": "Pro",
      "price": "49",
      "priceCurrency": "USD"
    }
  ]
}

Why SoftwareApplication and not Product? Both work, but SoftwareApplication is more specific and signals to AI engines that this is software you can sign up for, not a physical product.

The single biggest mistake we see: omitting the price field, or using a contact-us placeholder. AI engines treat missing price as missing information. Public price → cited. "Contact us" → not cited.

Article schema for blog posts

Article schema goes on every post. Here's the minimum:

{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "Post title",
  "description": "Post excerpt",
  "author": {
    "@type": "Person",
    "name": "Author Name",
    "url": "https://yoursite.com/about"
  },
  "publisher": {
    "@type": "Organization",
    "name": "Your Company",
    "logo": { "@type": "ImageObject", "url": "https://yoursite.com/logo.png" }
  },
  "datePublished": "2026-05-02",
  "dateModified": "2026-05-02"
}

Three details most sites miss:

  • Real author with url. Anonymous "FactSentry Team" works but a named author with a bio page is stronger.
  • dateModified updated when you edit. Models trust fresh content; updating the date when you do an actual update tells them.
  • Logo URL absolute, not relative. Models can choke on relative paths.

Organization schema, once, everywhere

Goes in your root layout so it ships on every page:

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Acme",
  "url": "https://acme.example",
  "logo": "https://acme.example/logo.png",
  "foundingDate": "2024",
  "sameAs": [
    "https://twitter.com/acme",
    "https://www.linkedin.com/company/acme"
  ]
}

The sameAs array is underrated. It links your site to your social profiles, which AI engines use to consolidate identity. List every public profile.

Validation

After you ship, validate. Two free tools:

  • Google Rich Results Test — pastes a URL and shows what Google parses.
  • schema.app validator — same, with a slightly more permissive parser.

If both show your schema parsing cleanly, you're done. If one or both throw errors, fix.

What to ship this week

Pick one page — your homepage. Add Organization and Product schema. Validate. Done in 30 minutes.

Next week: add FAQPage schema to any page with a Q&A block. Article schema to your top three blog posts.

Three weeks of patches and you're ahead of 70% of SaaS sites on structured data. After that, run a free audit and see how AI engines describe your brand now versus before.