Why it matters
Put Product schema on a hydraulic-pump spec page and the price, availability, MPN, and ratings still reliably earn rich results. That markup also hands an AI engine a clean entity to disambiguate, so it ties the part number to the right pump instead of guessing. FAQ schema is a different story. Google retired FAQ rich results on 7 May 2026, so it no longer earns a SERP chip, though the marked-up facts are still there for any machine that parses them.
Here is the honest part. Most AI answer engines, including ChatGPT and Perplexity, tend to read your JSON-LD as page text rather than as a validated graph. So the durable wins are the rich result and the entity clarity. Direct LLM ingestion of the schema is partly speculative, and we do not claim it as a guaranteed channel.
Proven value vs speculative value
Two buckets, kept separate so you spend effort where it pays off:
- Proven: Product rich results (price, availability, MPN) and entity disambiguation that any crawler can use.
- Speculative: an AI engine reading your JSON-LD as a structured fact source rather than as plain text on the page.
In practice
On a distributor SKU page for a Parker-style gear pump, mark up Product with mpn, gtin, offers, and aggregateRating, then validate it in Google's Rich Results Test. Keep the same facts visible in the page body and in a clean attribute table, because that text is what an AI engine most reliably extracts. Treat the schema as insurance for rich results and entity matching, not as a private API into the model.