Structured data is the difference between a page an assistant guesses at and a page it can parse as facts. Schema for AI helps Shopify merchants prepare Schema.org JSON-LD from store data you confirm. This page is the honest owner path: what we read, what you confirm, and which lifecycle states stay distinct. It does not claim that configured markup is already live, validated, or that it improves AI rankings.
Lifecycle states (keep these separate)
Public and in-app language must not collapse these into one claim:
| State | Meaning |
|---|---|
| Configured | You reviewed brand, policy, and catalog facts in AgentReady. |
| Generated | AgentReady produced JSON-LD (or metafield payloads) from confirmed inputs. |
| Theme extension enabled | The AgentReady theme app embed is on for the live theme. |
| Observed on storefront | A public fetch of your page shows the expected JSON-LD in the response. |
| Validated | Markup was checked against a validator or equivalent receipt for that page. |
Configured does not mean live. Generated does not mean the theme embed is on. Theme-enabled does not mean every page was observed. Observed does not mean validated. None of these imply ranking success.
What AgentReady reads from Shopify
On install and sync, AgentReady reads store data needed to score readiness and prepare schema, within the app's granted scopes. Typical inputs include:
- Shop identity (name, domain, contact surfaces Shopify exposes)
- Catalog: products, variants, collections, images, identifiers when present
- Prices, currency, and availability from live catalog state
- Policies and pages Shopify exposes (shipping, refund/return, privacy when present)
- Markets / localization signals the app is entitled to read
- Locations when LocalBusiness-style markup is relevant
Reading is not confirmation. Extraction is provisional until you confirm (see Product Law: never guess).
What you confirm
Before treating facts as customer-facing truth, confirm:
- Organization / contact — legal or trading name, logo, sameAs profiles, support contact
- Shipping — real rates/windows you stand behind (not invented)
- Return / refund — the policy customers actually get
- Privacy — accurate link and summary where schema references it
- Locations / markets — physical stores and market truth when claimed
- Catalog / product / offer truth — titles, identifiers, price, availability, variant facts
If a fact is missing or low confidence, AgentReady should ask rather than invent.
Expected owner path
- Install Schema for AI from the Shopify App Store. Free includes manual schema generation for up to 500 products and pages, with a six-hour cooldown; paid plans add automation and higher limits.
- Configure brand and policy facts; confirm provisional reads.
- Generate structured data for entitled page types from confirmed inputs.
- Enable the theme app extension (App embeds → AgentReady → Save) on the theme that serves customers. Until this is on, generated markup does not reach the live storefront.
- Observe a public product (or other) URL: view source or a fetch and look for AgentReady JSON-LD.
- Validate with a JSON-LD / rich-results tool when you need a receipt. Treat validation as a separate step.
Webhook-driven refresh can regenerate affected payloads after catalog or policy edits on plans with automation. Free supports manual generation, with a six-hour cooldown between runs. Regeneration still sits behind confirmation and theme delivery; it is not proof the public page is correct.
What types can be prepared
When data and entitlement exist, Schema for AI can prepare JSON-LD shaped for common AI-shopping and discovery needs, including:
- Product / ProductGroup, Offer (price, currency, availability, identifiers when real)
- AggregateRating only from real review data
- Organization / WebSite, BreadcrumbList
- FAQPage, HowTo, Recipe where they apply
- LocalBusiness when you enable and confirm location facts (paid plans)
Honesty rule: never invent ratings, prices, GTIN/MPN, or policies.
Theme app extension
The theme app embed is required for AgentReady schema to appear in the storefront HTML for shoppers and crawlers. It is designed to avoid customer-visible chrome and to remove cleanly on uninstall. Enabling the embed is a distinct merchant action from configuring or generating data in the app.
Troubleshooting (safe boundaries)
Missing JSON-LD on a live page
- Confirm the embed is enabled on the published theme, not only a draft.
- Confirm the plan's schema limits and generation mode. Free is manual, with a six-hour cooldown; paid plans add automation.
- Confirm generation ran for that resource after your last catalog change.
- Soft-cache / CDN: re-fetch with cache bypass after saving the theme.
- Other apps or theme sections may omit or replace head output; compare raw HTML.
Duplicate JSON-LD
- Themes, review apps, and SEO apps often emit a second Product node.
- Prefer one coherent graph. Disable overlapping emitters or keep AgentReady as the structured-data owner where you choose it.
- Duplicates are a quality problem; they are not evidence of ranking harm or benefit by themselves.
Stale price or availability
- Confirm the product was updated in Shopify, then confirm AgentReady regenerated, then re-observe the public page.
- Do not equate "webhook received" with "storefront observed."
If you need help, say which state you reached (configured / generated / theme-enabled / observed / validated) and the public URL. Do not expect support to claim AI ranking outcomes.
What it does not claim
- AgentReady does not guarantee ChatGPT, Perplexity, or any assistant will recommend you because schema exists.
- It does not silently mutate catalog copy without confirmation.
For discovery files that help crawlers find your data, see agents.md and llms.txt. For measuring answers over time, see AI answers.