FAQ Schema, formally the FAQPage structured-data type, is JSON-LD markup that tells search and AI engines a page contains a set of questions with direct answers. Google deprecated the FAQ rich result on May 7, 2026, but AI engines still extract from FAQPage markup when deciding what to cite.
Across UC's 146-article Australian corpus audit, 35% of audited sites had a visible FAQ block in the body but only 24% wired it up with FAQPage JSON-LD, roughly a third of FAQs sit unparseable to the AI engines they're meant to feed.
That gap is the reason the markup still matters. Before May 2026, FAQ schema had a visible payoff: Google would expand the questions as collapsible links under your search result, giving your listing more real estate. Google's own deprecation notice is blunt: "As of May 7, 2026, FAQ rich results are no longer appearing in Google Search." The visual snippet is gone. What remains is the structured signal underneath it, and that signal still reaches every AI engine that reads the web.
How the type works. Schema.org defines FAQPage as a web page presenting a set of frequently asked questions and answers. The implementation is JSON-LD: a FAQPage entity containing a list of Question entities, each with an acceptedAnswer carrying the answer text. A crawler or AI extraction system reading that block gets the Q&A in machine-readable form, without parsing prose. Schema.org was founded by Google, Microsoft, Yahoo and Yandex as a shared vocabulary for exactly this, not just for classic search, but for any system that needs to understand what a page contains, AI engines included.
The practical implication: when Perplexity or ChatGPT answers a question, it pulls from pages whose Q&A structure it can parse cleanly. FAQPage markup is one of the cleaner parsing targets, unmarked prose paragraphs are harder to extract with confidence. The deprecation of Google's rich snippet is not a signal to remove FAQ markup, it's a signal that the audience for the markup shifted from Google's SERP renderer to the AI layer on top.
What belongs on a page with FAQ schema. Acceptable uses are real Q&A pages where the content genuinely answers questions the audience asks. Pages with user-submitted answers, promotional copy, or FAQs duplicated across the whole site are the wrong candidates. One specific FAQ block per page, tied to that page's topic, is the shape that earns extraction, from AI engines now and from Answer Engine Optimisation signals broadly. What works in practice: a buyers agent's service page with five real questions buyers ask, how does pricing work, do you cover Melbourne's eastern suburbs, what happens if the property fails the building report, each marked up with acceptedAnswer JSON-LD, none lifted from a generic FAQ template. That's the version Perplexity quotes.
FAQ schema sits inside the wider schema markup system and feeds directly into SEO & AI Visibility work. The rich snippet is dead; the structured answer signal is not.