SameDayDesk · Guide/Comparison · June 2026

How to add JSON-LD structured data for AI search (2026)

Paste an Organization block and an Article block into your <head>. That is the machine-readable layer ChatGPT, Perplexity and Google AI Overviews actually extract. Copy-paste examples below, plus the SaaS benchmark that shows 1 in 3 top sites ship none of it.

1 in 3top SaaS sites ship no JSON-LD at all (our 24-company scan)
8 / 24SaaS leaders with zero structured data: Notion, Linear, Airtable, Clerk and more
~32%visibility lift from adding statistics; +41% from quotations (GEO paper, KDD 2024)
~4.4xAI-search referral traffic converts vs organic search

The direct answer: two blocks, in the head, today

JSON-LD is a small JSON script that describes what a page is in a format machines read without guessing. For AI search in 2026 you need two blocks on the page, server-rendered into the <head> so a crawler that does not run JavaScript still sees them:

  1. Organization — who publishes the site (name, URL, logo, social profiles). One block, sitewide.
  2. Article — what this specific page is (headline, author, publisher, dates). One block per content page.

Here is the Organization block. Drop it in the <head> of every page and change the four values:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Your Company",
  "url": "https://yourdomain.com/",
  "logo": "https://yourdomain.com/logo.png",
  "sameAs": [
    "https://twitter.com/yourhandle",
    "https://www.linkedin.com/company/yourcompany"
  ]
}
</script>

And the Article block, one per content page:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "Your exact page title",
  "datePublished": "2026-06-24",
  "dateModified": "2026-06-24",
  "author": { "@type": "Organization", "name": "Your Company" },
  "publisher": {
    "@type": "Organization",
    "name": "Your Company",
    "logo": { "@type": "ImageObject", "url": "https://yourdomain.com/logo.png" }
  },
  "mainEntityOfPage": "https://yourdomain.com/this-page-url"
}
</script>

That is the whole job for most pages. Validate it once at validator.schema.org, ship it, and the parts AI engines care about are in place. The rest of this page explains why these two types and not the ones you were probably about to reach for.

Skip FAQPage and HowTo for "rich results" — that win is gone

If your old SEO checklist says "add FAQPage schema for the accordion in search," delete that line. Google deprecated and removed FAQPage and HowTo rich results between 2023 and 2026. The markup is still valid and still machine-readable Q&A, so you can keep a small FAQPage block as a secondary signal — but it will not produce a SERP feature, and you should not build your structured-data strategy around it.

The durable, extractable layer is Organization and Article. For a free tool, add SoftwareApplication. For a paid offering, add Service with an Offer that carries a real price. Those are the types that survive, and they map cleanly onto what an AI engine needs to attribute a claim: who said it, on what page, published when.

Schema type2026 statusUse it for
OrganizationCore — extractedPublisher identity, sitewide
ArticleCore — extractedEvery content / blog page
SoftwareApplicationCore — extractedA tool or app page
Service + Offer (price)Core — extractedA paid product / service
FAQPageRich result removedOptional secondary Q&A only
HowToRich result removedDo not rely on it

Why this moves the needle for ChatGPT, Perplexity and AI Overviews

AI search retrieves live. ChatGPT Search and Microsoft Copilot largely pull from the Bing index — about 87% of SearchGPT citations matched Bing's top 20 (Seer Interactive, reconfirmed by Search Engine Land, April 2026). Perplexity runs its own crawler and tends to cite fresher pages, so first citations on low-competition queries can land in days. Google AI Overviews is the wild card: citation there is largely independent of organic rank, with about 68% of AIO-cited pages not in the top 10 organic results.

Because retrieval is live (RAG), a freshly indexed page can be cited without ever being in a model's training set. That is the opening. But to be cited, the engine has to parse your page cleanly — and that is exactly where structured data earns its keep. JSON-LD hands the machine the headline, the publisher, and the dates as data instead of asking it to scrape them out of your markup.

About 44% of LLM citations come from the first 30% of a page, and adding direct quotations raised visibility about 41% in the Princeton/Georgia Tech GEO study (KDD 2024). Put your answer up top, mark up who said it, and you are playing both sides of how these systems pick sources.

The benchmark: the sites you'd assume are dialed in, aren't

We scored 189 well-known companies across 10 industries on six fundamentals, including JSON-LD structured data. SaaS averaged 87/100 — respectable — yet 8 of the 24 SaaS sites we scanned ship no JSON-LD at all. That list includes Notion, Linear, Airtable, Clerk, Cal.com, PostHog and Gumroad. These are product-led teams that obsess over conversion, and they are leaving the machine-readable layer blank.

The contrast is sharp. Stripe, Supabase, Webflow, Vercel (93) and HubSpot (90) score an A. Meanwhile Figma (73), Linear (71), Substack (68), Airtable (68) and Clerk (61) sit in C territory. The gap between an A and a C here is rarely talent — it is whether someone pasted two blocks into the <head>.

CompanyScoreJSON-LD
stripe.comAYes
vercel.com93 (A)Yes
hubspot.com90 (A)Yes
figma.com73 (C)Limited
linear.app71 (C)None
airtable.com68 (C)None
clerk.com61 (C)None

It is not just SaaS. In the full 189-company set, OpenAI and GitHub scored a D on JS-heavy homepages with thin server-rendered content; Perplexity — an AI search engine — scored a C; LlamaIndex scored a D; Klarna scored an F (38, the lowest). Ars Technica scored an F too: it lets crawlers in but ships no structured data. The full raw data is open under CC-BY: ai-search-readiness-2026.csv.

Check your site in 30 seconds

Run the free browser checker to see whether your JSON-LD, crawler access, Open Graph and sitemap pass — the same six fundamentals we scored 189 companies on. Then grab the $9 Kit with every template ready to paste.

Run the free check Get the $9 AI Readiness Kit

What to do this week

  1. Add the two blocks. Paste Organization sitewide and Article on every content page, server-rendered into the <head>. If your homepage is a heavy JavaScript bundle (the trap OpenAI and GitHub fell into), make sure the JSON-LD is in the initial HTML, not injected after load.
  2. Validate. Run it through validator.schema.org. Zero errors, no exceptions.
  3. Tool and product pages. Add SoftwareApplication to any app page and Service + Offer (with a real price) to anything you sell.
  4. Push to the index now. Bing indexes new content in hours to days, and IndexNow needs no account — that is the fast lane to ChatGPT visibility. Do not gate on Google: new domains can sit in a 3–9 month sandbox on commercial queries.
  5. Skip llms.txt as a priority. Google's Gary Illyes says it is not supported and not planned, and roughly 97% of LLM crawler hits never fetch it (Ahrefs). Optional hygiene, not a ranking factor.
  6. Audit it. Run our open-source CLI from your terminal: npx github:epistemedeus/ai-readiness yoursite.com (repo).

Want it done for you? The $39 Fix Pack ships the exact JSON-LD, Open Graph and crawler config built for your site, same day. And the $249 AI-Search Visibility Audit tests whether ChatGPT, Perplexity and AI Overviews actually cite you against named competitors — the thing structured data is supposed to unlock. Given that AI-search referral traffic reportedly converts about 4.4x organic, and comparison and "alternatives to" formats draw roughly 40.9% of AI citations on commercial-intent queries, this is the cheapest visibility you will buy this quarter.