FAQ Schema: What It Still Does in August 2026
Google shut off the FAQ rich snippet in 2026 — here's why the underlying structure still matters for AI answer engines anyway.
This guide reflects Google's own published documentation and changelog entries as of August 2026, including a feature removal (FAQ rich results) that took effect in May 2026. Verify current status directly against schema.org and Google Search Central before treating any deprecation date here as still accurate.
TL;DR — Google restricted FAQPage rich results to government and health sites in August 2023, then removed the feature from Google Search entirely as of May 2026 — it even deleted its own implementation documentation. That doesn't make FAQPage schema pointless: it's still valid schema.org vocabulary in active use on millions of domains, and clean question-and-answer structure is exactly the shape AI answer engines look for when extracting a direct response — the payoff just moved from a Google SERP rich snippet to broader AI-answer eligibility.
What FAQ schema is
FAQPage is a schema.org structured-data type for marking up a page's frequently-asked-questions content — each question and its answer gets wrapped in machine-readable markup, written as JSON-LD, so a system reading the page doesn't have to guess where a question ends and its answer begins. Google introduced support for it as a visual rich result in May 2019.
What actually happened to it
In August 2023, Google restricted FAQ rich results to "well-known, authoritative" government and health websites, after widespread abuse — sites were stuffing pages with artificial FAQ sections purely to grab extra space on the results page, often with questions nobody was actually asking. Then in May 2026, Google removed the FAQ rich result feature from Search entirely, for every site, government and health included — and by June, Google had deleted its own FAQPage implementation documentation, stating plainly that the feature "is no longer shown in Google Search results."
So if a guide anywhere still promises FAQ schema will get you a rich snippet on Google, it's describing a feature that's been gone since mid-2026.
Why it's still worth doing
The schema.org FAQPage type itself didn't go anywhere — schema.org's own usage data shows it's still deployed on somewhere between 1 and 10 million domains as of mid-2026, well after Google stopped rewarding it with a visual SERP feature. That gap is the actual reason it's still worth using: the value was never really the markup itself, it was the discipline of writing a clear question paired with a direct, self-contained answer — exactly the shape an AI system looks for when deciding what to extract into a generated response, independent of whether Google renders a special box for it.
In other words: the Google-specific reward disappeared, but the underlying content pattern FAQ schema forces you into — one real question, one direct answer, no padding — is still doing useful work for every other AI system reading the page.
How to actually implement it now
Write it in JSON-LD (the format schema.org and every major consumer prefers), with one FAQPage object containing a mainEntity array of Question items, each carrying its own nested acceptedAnswer. Keep every question genuine — something a real reader or an AI system would plausibly ask — and every answer complete enough to stand alone if it's the only sentence an AI system decides to quote.
Skip the temptation to mark up decorative or SEO-bait questions purely to add more entries; that's the exact behavior that got the feature restricted for everyone in the first place, and a system extracting for a genuine answer will simply skip content that doesn't read like one.
- One FAQPage object per page, not one per question — the questions nest inside it as a mainEntity array.
- Use real questions a person would actually type or ask out loud, not keyword-stuffed variants of the same query.
- Write each answer to make sense on its own, out of context — that's the same standard an AI system extracting it has to meet.
- Validate the markup against schema.org's FAQPage spec directly, since Google's own Rich Results Test no longer checks for it.
Where to go from here
What Is AEO covers the broader shift this reflects — content structured for extraction rather than for a specific SERP feature. The FAQ Extractor concept page covers tools built specifically to pull genuine FAQ content out of existing pages instead of writing it from scratch.