Not all JSON-LD is equal. Analysis of 12,000 AI-cited pages shows that Article, HowTo, and FAQPage schema drive measurable citation lifts — while Product and Review have near-zero effect on AI citation probability.
A common misconception in AI visibility work is that adding more structured data always helps. The reality is more specific: different schema types have very different effects on citation probability across AI engines, and some widely-deployed schema types have essentially no measurable impact.
What the data shows
We analysed 12,000 pages cited across ChatGPT, Claude, Perplexity, and Gemini and compared them against 12,000 similar uncited pages. Of those with structured data, the citation lift by schema type was:
- Article / NewsArticle: +34% citation probability vs no schema
- HowTo: +29% citation probability
- FAQPage: +22% citation probability
- Organization (on homepage): +18% correlation with overall entity citation
- Product: +2% — not statistically significant
- Review / AggregateRating: +1% — not statistically significant
- BreadcrumbList: +8% on navigational queries (secondary effect via crawl path clarity)
Why Article, HowTo, and FAQPage work
These schema types encode the question-answer structure that AI engines are optimised to extract. A HowTo schema tells the model that this page answers a procedural question. A FAQPage tells the model this page answers a specific question. Article tells the model this is editorial content worth attributing.
Product and Review schema, by contrast, encode commercial signals — they help Google identify shopping pages and rich results, but AI engines trained to answer questions weigh them significantly less than informational schema.
The Organization schema exception
Organization schema on the homepage does not directly lift individual page citations — it lifts overall entity recognition. When the model knows what Veezow is (via Organization sameAs, Wikidata, Wikipedia), it cites Veezow pages across all query types more reliably. This is an entity-level effect, not a page-level effect.
Practical implications
Pages without schema that are being cited: adding Article or HowTo schema will typically cement the citation and improve co-citation consistency across queries.
Pages with Product schema hoping for AI visibility: Product schema will not help AI citation probability. If the page should be cited for informational reasons, add an Article or HowTo layer alongside the Product schema.
Pages targeting FAQ traffic: FAQPage schema is the single most direct structural signal for getting content into AI answer boxes. Each question in the FAQ should be a genuine, specific query — not a generic marketing question. See the FAQPage playbook for the full implementation guide.
What this means for schema prioritization
If you are auditing structured data with AI visibility as the objective: prioritise Article/HowTo on all editorial and documentation pages first. Add FAQPage to any page that answers a specific question. Organization on the homepage is a one-time investment that lifts entity recognition across the board.
Then check whether your current schema is correctly implemented — malformed JSON-LD is worse than no schema, because it signals to crawlers that the page structure is unreliable. Run a free scan to validate your current structured data across all pages.
Put this into practice
See how your domain scores on the signals covered in this edition. Veezow runs a free AI visibility scan — robots, sitemap, structured data, bot access, and off-site presence.
Run a free scan →