Schema markup recommendations and AI enhancement

Every audit includes schema markup recommendations: for each crawled page, Daylytix works out the page type and writes ready-to-copy JSON-LD. An optional AI step can then enrich the markup with details from the page itself.

Where to find it

Open an audit and go to the Schema Markup tab. The Schema Recommendations table lists each page with its detected type and a status:

• Missing - the page has no schema; copy the generated code and add it.
• Incomplete - schema exists but is missing fields, which are listed.
• OK - the existing schema looks good.

Use the filter box to find a URL or a type. Click a row to see the code.

How the page type is chosen

The recommendations are generated from the crawled data, without AI. Daylytix checks, in this order:

1. Legal and utility pages (privacy, terms and similar) always get a plain WebPage.
2. Schema already on the page: a specific schema.org type found there is kept. Article markup counts only if it has an author or date, or sits on a blog URL.
3. The page’s og:type, and whether it is the homepage.
4. URL patterns, for example /blog/, /product/, /category/, /contact, /about (Organization) or /location (LocalBusiness). A /faq URL also needs at least two question-and-answer pairs.
5. Hints in the title for recipes, events and FAQs, and a price in the page data for products.
6. Three or more question-and-answer pairs on the page suggest FAQPage.

Connected graph or single block

The code view offers two formats:

• Connected @graph (recommended) - one block that links Organization → WebSite → WebPage → the page’s main content, so search engines see how they relate.
• Single block - just the markup for the page’s own type.

Both can be downloaded for all pages as CSV from the top of the table. The site search box markup is not added, because Google retired the sitelinks search box in 2024.

Enhance with AI

Click Enhance with AI on a page to have the markup extended. Daylytix fetches the live page and gives the AI its real text, images and any existing JSON-LD. The AI may add properties and choose a more specific type, but only from facts found on the page. Its answer is then checked: properties that don’t exist in schema.org and URLs that are not on the page are removed. It uses your Gemini key first and your Anthropic key as a fallback - see Setting up your AI key.

Adding the code to your site

The copied code is the JSON-LD itself. Wrap it in a script tag and place it in the page’s <head> (the <body> works too):

<script type="application/ld+json">
  ...the copied code...
</script>

• WordPress - use a plugin or your theme’s header settings that allow custom code in the head.
• Google Tag Manager - a Custom HTML tag with the script, fired on that page.
• Templates - for many pages of one type, add the markup to the template with the page’s fields filled in dynamically.

Validate before you rely on it

Test the page in Google’s Rich Results Test and the Schema.org Validator. Valid markup makes a page eligible for rich results, but Google decides when to show them - and FAQ rich results are now shown only for well-known government and health sites.