Breadcrumb schema helps AI answers indirectly — it clarifies a page's position in the site's information hierarchy, which improves classic rich-result eligibility and gives answer engines a cleaner path-to-topic when they attribute a citation.
The type is BreadcrumbList, documented for Google Search in the breadcrumb structured-data guide. Each entry in the list is a ListItem with a position, a name, and an item URL.
Two rules make the markup useful. First, the visible page should render a breadcrumb trail that matches the JSON-LD — Google's structured-data guidelines require visible and marked content to align. Second, each item URL should be the canonical URL of the page it points to.
For AI answers, the practical benefit is contextual: when an engine cites site.com/glossary/qapage, a breadcrumb makes it obvious that the page sits under a "Glossary" section of a site about AEO. That context propagates into how the answer describes the source.
Include a BreadcrumbList on every deep page — glossary entries, answer pages, case studies — and skip it only on the homepage and top-level index pages where there is no path to display.