<b>The headings Google rendered but refused to weigh</b>

<b>The headings Google rendered but refused to weigh</b>

A design-tool company moved their docs to web components. Pages rendered flawlessly, indexed fine — but ranked for almost nothing despite thousands of pages of real content.

The rendered HTML in URL Inspection looked complete. That's what threw us. The content was visibly there.

We extracted the indexed text Google actually associated with each URL using exact-phrase site: searches. Most in-content phrases returned nothing. Google had the DOM but wasn't crediting the text.

Their headings and body lived inside Shadow DOM, encapsulated by custom elements. Google renders Shadow DOM into the flattened tree for the snapshot — but the team had used closed shadow roots with content injected via <code>slot</code> from inline templates that loaded after first render. The flattened tree captured at snapshot time held empty slots.

Closed shadow roots plus late slot assignment is a content-invisibility recipe even when the page looks perfect to humans.

Fix: open shadow roots, slotted content present in initial server HTML.

Result: indexed-phrase coverage went from 12% to 94%, impressions up 6x in nine weeks.