People ask chat tools whether a product is waterproof, and the product page often never says. A useful product FAQ gives a direct answer a person can quote: the fact first, then the condition. Product FAQs for AI Crawler stores those questions and answers in Shopify metafields and emits FAQPage JSON-LD in the light DOM, outside the widget's shadow DOM, built from that FAQ data. That makes the text available without running the widget script.
None of that is a promise that an assistant will cite the page, recommend the product, or show a rich result. Write the answer because a shopper on the page needs it. Treat any reuse by a search or chat tool as outside the app's control. The feature catalog documents the markup. It does not document a ranking, a citation rate, or an AI overview.
A quotable answer
An answer engine can only repeat what the page actually says. "Is the shell waterproof?" needs a sentence such as "The shell is not waterproof. It sheds light rain and should not be worn in a downpour." A paragraph about adventure and craft gives the tool nothing to quote, so it may guess. Guessing is how a product gets described as waterproof when it is only water resistant.
Lead with the fact. Name the unit, the material, or the limit. Then add the condition: which size, which wash, which country you ship from. Keep the answer on the product when it would be false for another item. A shop-wide group is for a rule that is true across the catalog. Assistants that read one URL will treat the text on that URL as the product's answer, so the wrong scope creates the wrong quote.
- Fact first: waterproof or not, the measurement, the fiber, the ship window.
- Condition second: the size, the wash method, or the region the rule applies to.
- Source on the page: the same words a support agent would send, not a slogan.
What the app stores
When you save a group, the questions and answers are written into Shopify metafields. The product-page block reads the product's FAQ metafield. The page block reads a shop-wide group that has no products assigned. The storefront does not call the app backend to display the list. It does call the app to log analytics. Display and analytics are separate.
Both blocks emit a real FAQPage JSON-LD script in the light DOM. The main entity is the list of questions and accepted answers, taken from the FAQ data. Question and answer text is stripped of HTML and JSON-escaped before it is embedded. The widget itself renders in a shadow DOM, using Accordion, Grid/cards, Boxed, or Minimal. The structured data sits outside that shadow root on purpose, so a crawler does not have to execute the widget to read the pairs.
AI generation can draft those pairs from the product title, description, vendor, and type, or from a description you write, with clarification questions in description mode. Drafts are editable before you save. If the description never said the shell was waterproof, do not save a draft that claims it was. A crawler will treat the saved sentence as the page's answer.
What this does not guarantee
The catalog does not say that Google, or any assistant, will quote the FAQ, expand it in search results, or prefer the product because the JSON-LD exists. Visible text and structured data can disagree if another app or the theme prints a second copy. Check the live page source for the FAQ text you saved. If two systems describe the same product differently, fix the words before you worry about markup.
A blog post about the product is a different URL from the product page. An assistant that answers "is this jacket waterproof?" should be reading the product, not a general article. Keep the definitive sentence on the product. Use the blog, if you write one, to explain a topic, and do not paste FAQPage JSON-LD into the article body. Article pages on this theme already have their own article structured data.
A good answer and a weak one
Weak: "Our jackets are built for whatever the day brings, with quality materials and a fit you'll love." That sentence has no fact. A tool cannot tell a shopper whether to pack it for rain.
Stronger: "This jacket is water resistant, not waterproof. Light rain beads on the face fabric. Seams are not taped, so a sustained downpour will soak through." A shopper can act on that. An assistant that quotes the page has a limit to repeat. You can still be wrong if the factory changes the fabric and you do not edit the FAQ. Saving the group again rewrites the metafield, which is how the page catches up.
Keep that sentence on the product URL shoppers and assistants are most likely to open. A homepage banner and a social caption are easy to forget when the fabric changes. The product FAQ is the copy you assign to that product and edit in one group. When the spec changes, edit the group and save so the metafield and the light-DOM FAQPage stay aligned with the words beside add to cart. The App Store listing is Product FAQs for AI Crawler.
Questions about FAQs and AI answers
Does the app emit FAQPage JSON-LD?
Yes. The product block and the global block emit it in the light DOM, outside the widget shadow DOM, from the saved FAQ data.
Will that get the product cited in chat tools?
The catalog does not promise citations, recommendations, or rich results. Write the answer for the shopper on the page.
Should the answer live only in an image?
No. A size chart that is only a picture gives text tools nothing to read. Put the measurement in the FAQ text.
Read how the metafield and the light-DOM FAQ markup fit the product page on Product FAQs for AI Crawler. Put one quotable sentence on the product people already ask assistants about, and delete any draft that invents a spec.