The FAQ that earned a rich result, then lost it

The FAQ that earned a rich result, then lost it

A software docs site had FAQ rich results on 80 pages — extra SERP real estate, great CTR. A refactor wiped them out in two weeks. The FAQ content was still right there on every page.

The data: 80 rich results dropped to zero. Structured-data reports showed 'no items detected' on pages that visibly had FAQs.

The investigation went to how the FAQ section mounted. The refactor wrapped it in an Intersection Observer for performance — the component, and its JSON-LD structured data, only rendered into the DOM when the section scrolled near the viewport. Google's renderer doesn't scroll to the bottom of a long doc page, so the FAQ and its schema never entered the rendered HTML it evaluated.

Turns out attaching structured data to a viewport-triggered component is the same as not having structured data at all. The markup that earns rich results has to exist before any scrolling.

The fix: render the FAQ content and its JSON-LD into the initial HTML, and reserve the Intersection Observer for genuinely below-the-fold images, not content or schema.

Result: rich results returned to 78 of 80 pages within three weeks, and FAQ-driven CTR recovered the ~22% it had quietly lost.