Jobs to be done · Fix · SEO · Dev

How to fix product schema for AI search.

Your Product JSON-LD is a mess, you don't know which fields the engines actually weigh, and the schema app you installed filled the easy fields and called it done. Hand-authoring is correct and unscalable; apps are scalable and shallow. The job needs both properties at once.

Quick answer

Run the schema audit in eCommerce Insights, review the field-level gap report, approve the generated diffs per SKU, and push to Shopify metafields (early access, Growth plan). For a one-off SKU, the free Product Schema Generator drafts complete JSON-LD from any product URL.

The slow way: validator, theme liquid, repeat

The manual loop: open a PDP, view source, paste the JSON-LD into a validator, list what's missing, write a fix. By SKU forty you hand it to a developer, who opens the theme liquid, discovers the block is half metafield-driven and half hard-coded, and estimates three weeks. The alternative — install a schema app — produces a more elaborate block that validates cleanly and still doesn't move citations, because it fills the derivable fields and ignores the ones that matter.

The underlying problem: schema is a content problem, not just a compliance problem. material, origin, fabric, and care values come from the brand's knowledge of its own catalog; aggregateRating comes from a review app that never talks to the theme. No generic tool can fill those — and no validator will tell you they're the fields that matter. The field-by-field reference is in the schema for AI search guide; the canonical type definition is at schema.org/Product.

The eCommerce Insights way

Audit the live schema. One click on a connected store. eCommerce Insights fetches every PDP, parses the JSON-LD the theme actually emits — not what the app claims — and validates against schema.org plus the field expectations derived from AI citation data.
Review the field-level gap report. Gaps group by type: aggregateRating missing on 312 SKUs, material empty on 95, gtin malformed on 12. Batch-fix by gap, not by product.
Generate complete JSON-LD per SKU. Drafts pull from product JSON, metafields, and review-app data. Values the system cannot know — origin, fiber blend — are flagged for a human with the right Shopify metafield named.
Approve the diffs. Every proposed block arrives as a diff against the live version, reviewable by a merchandiser without reading raw JSON. Approve hero SKUs individually, long-tail in batch, reject anything that needs editing.
Push to Shopify metafields. Approved schema writes to a dedicated namespace the theme references — no theme-file edits after initial wiring. Shopify push · Early access CSV and snippet export on every plan. Definitions in the product schema glossary entry.

What "good" looks like

Core 11 fields present (name, description, sku, gtin, brand, offers, image, material, color, aggregateRating, review)95%+ of PDPs

PDPs below 60 on the structured-data factor0

aggregateRating wherever review data exists100%

Rich Results Test errors on hero SKUs0

Field-citation correlations from eCommerce Insights audit data as of early 2026. The most common single miss across Shopify stores: review data living in an app, never reaching the JSON-LD. Fixing that one wire is frequently worth more than everything else on the gap list combined.

Ask AI about this job

Have your favorite AI engine apply this walkthrough to your catalog.

Frequently asked questions

Which Product schema fields actually matter for AI search?

Eleven fields correlate most with AI citation in eCommerce Insights audit data as of early 2026: name, description, sku, gtin, brand, offers (price, availability, url), image, material, color, aggregateRating, and review. The single highest-leverage addition on most Shopify PDPs is aggregateRating — the review data usually exists in an app and never reaches the JSON-LD.

My theme already emits Product JSON-LD. Isn't that enough?

Usually not. Most themes emit a technically valid but structurally shallow block — name, price, image — and skip the metafield-backed fields (material, gtin, color) and the review aggregate. The audit reads what your theme actually outputs and shows which fields are empty, generic, or absent per SKU.

Can eCommerce Insights push schema back to Shopify?

Yes, in early access on the Growth plan. Approved schema writes to Shopify metafields under a dedicated namespace, and the theme references those metafields in its JSON-LD block — no theme-file edits after the initial wiring. Every push is logged and reversible. Everyone else exports snippets or CSV.

Why didn't my schema app move my citation rate?

Because schema is a content problem wearing a technical costume. Generic apps fill the fields derivable from product JSON and stop. The fields engines reward — material, origin, dimensions, review aggregate — come from your catalog knowledge and your review app, which is exactly what the diff workflow is built to capture.

How often should I re-audit product schema?

Continuously on a paid plan — theme updates, app changes, and new launches each break schema in specific ways, and the scan catches regressions within a refresh cycle. Doing it manually, monthly is a reasonable baseline, with the free Product Schema Generator for spot checks.

Related jobs

Fix

Schema fixed as diffs, not tickets.

Field-level gaps across the catalog, drafts a merchandiser can approve.

Generate schema for one product Read the schema guide