Schema.org releases new FAQPage recommendations for AI-driven search — what merchants need to change

Schema.org released an updated FAQPage specification on July 28, 2026, with explicit guidance on how structured FAQ content should be authored for consumption by large language models and AI-driven search systems. The update is the first time Schema.org has directly addressed LLM citation behavior in its usage recommendations.

1. Anchor linking is now recommended, not optional. The new guidelines strongly recommend that each FAQPage entry include a unique @id with a corresponding URL fragment anchor on the page. Previously, this was considered a best practice; it is now part of the core recommendation. The rationale: LLM citation systems need a verifiable page-level reference point for each answer.

2. Answer text must be in the DOM on initial page load. The spec now explicitly warns against JavaScript-dependent answer rendering. If the acceptedAnswer text is loaded dynamically, LLM crawlers may not see it. The recommendation is to use server-side rendered or static HTML for answer content.

3. Question granularity carries extraction weight. The update notes that broad questions (e.g., "What is your return policy?") are less likely to be cited by AI systems than specific questions (e.g., "What is your return window for international orders?"). This is the first time Schema.org has suggested that question specificity affects machine extraction likelihood.

Visora's audit database, covering over 25,000 ecommerce product pages, shows that only 27% of current FAQPage implementations pass all three of the new recommendations. The most common failure point is missing anchor links — 68% of scanned pages have no @id on their FAQPage entries. The second most common is answer content loaded via JavaScript toggle — present in 41% of pages.

For cross-border ecommerce stores, the update has immediate practical implications:

• FAQPage implementations that lack anchor links may see a gradual decline in AI citation frequency as LLMs update their retrieval systems to prefer anchored content.
• Stores with JavaScript-dependent FAQ expand/collapse widgets need to migrate to static HTML or server-side rendered content to maintain citation eligibility.
• Writing product-specific, market-specific FAQ questions rather than generic templates now has measurable citation value.

Run your store's top 10 product URLs through Visora's free audit at geovisora.com/audit. Check the Schema Completeness score. If it is below 70, review the FAQPage section specifically for anchor linking and answer visibility issues. The fix for most stores takes under an hour per page and produces a measurable citation signal improvement.