The JSON-LD Implementation
Add FAQPage schema to every page with a FAQ section. Show in a dark styled code block:
Place this in the page's @graph block alongside the WebPage and Speakable entries, not as a standalone script tag.
Writing FAQ Content That Gets Cited
The question phrasing determines which queries the FAQ pair is a citation candidate for. Match the exact language users use in AI chat interfaces.
Strong phrasing
"How much does a roof replacement cost in Phoenix?" — matches the natural question format users type or speak.
Weak phrasing
"Roof Replacement Pricing" — a heading, not a question. AI systems do not extract headings as FAQ questions.
The answer must begin with the direct response in the first sentence. No preamble. No "great question." The AI extracts the text of the acceptedAnswer field and may truncate after the first 100 to 150 words, make the first sentence count.
Include at least one specific, verifiable fact in each answer. A range, a timeframe, a specification, a code reference. Generic answers are not citation candidates.
FAQ Schema and Traditional SEO Rich Results
Before AI Overviews, FAQ schema was used to generate rich results in the standard organic listings, expandable FAQ dropdowns visible in the search results page. Google has scaled back these rich results since 2023, but the underlying markup still improves AI citation probability. Building FAQ schema today optimizes for both the current AI citation use case and any future reintroduction of rich results.
Google first used FAQ schema for rich results. The same markup now powers AI Overview extraction — legacy investment, compounding returns.
FAQ Section
Five is the minimum for AI citation purposes. Eight to twelve is optimal, enough to cover the full range of questions a potential customer might ask, without becoming so long the page loses focus. Each FAQ pair should address a distinct question with no overlap.
As closely as possible without being unnatural. If users commonly search "how much does HVAC replacement cost," write the FAQ as "How much does HVAC replacement cost?", not "What are the investment considerations for a new HVAC system?" The closer the match to natural query phrasing, the higher the extraction probability.
Use it on every page that contains genuine question-and-answer content, service pages, pillar pages, location pages, blog posts. You do not need a dedicated FAQ page to use FAQ schema. In fact, distributing FAQ schema across all relevant pages is more effective than concentrating it on one page.
Yes. Voice assistants use FAQ schema as a primary extraction source for spoken answers. When someone asks their phone's assistant "how long does roof replacement take," the assistant's answer frequently comes from an FAQ schema-marked answer rather than continuous body copy.
Google's Rich Results Test will flag errors in the schema, missing required fields, incorrect property types, or mismatched entity references. Errors cause Google to ignore the structured data entirely, which means the FAQ pairs provide no extraction signal. Validate all schema before publishing.
Five Questions Per Page. Five Additional Citation Opportunities. The Math Is Simple.
Every service page on your website should have at least five structured FAQ pairs, each in schema markup, each written to match real user query phrasing. SEO Local writes and implements this on every page we build.
Get Started
SEO Local