The crawling process
When an AI engine crawler visits your website, it reads the full HTML of each page, including any `<script type="application/ld+json">` tags in the `<head>` or `<body>`. These tags contain your JSON-LD structured data — the schema.
The crawler parses the JSON-LD, extracts the entity declarations, validates them against the schema.org type definitions, and feeds the valid structured data into the engine's knowledge pipeline.
This process is automatic. You do not need to submit your schema separately — it is discovered as part of normal crawling.
The critical requirement: server-side rendering
The most common schema implementation error that causes crawlers to miss structured data is client-side rendering. Some website platforms (React, Vue, Angular applications) render page content — including schema — using JavaScript that executes in the browser after the initial HTML is served.
Most AI engine crawlers process the initial HTML but have limited JavaScript execution capability. Schema embedded only in client-side rendered JavaScript may never be crawled.
The solution: Ensure your JSON-LD structured data is included in the server-side rendered HTML — the raw HTML that the crawler receives before JavaScript executes. In Next.js, this means your schema should be in the page's server-rendered output (not added only via useEffect or similar client-side code). In WordPress, schema plugins that inject into the `<head>` are generally server-side rendered.
Accelerating discovery
After adding or changing schema, submit the affected pages for indexing in Google Search Console (URL Inspection tool → Request Indexing). This does not guarantee immediate crawling, but signals to Google that the page has changed and accelerates re-evaluation.
Perplexity and ChatGPT do not offer direct indexing submission tools — they crawl on their own schedules.
Verifying discovery
After 2–4 weeks, check Google Search Console's Rich Results Status report to confirm your schema has been crawled and validated. If schema is not appearing in the report, it may not be reaching the crawler — investigate your rendering pipeline.
