Product feed vs Product schema: two copies of the truth that must agree

An ecommerce catalogue usually tells the same product story in at least three places: the page a shopper sees, the structured data inside that page and the feed sent to Merchant Center. They are often owned by different systems. That is how a $79 product becomes $74 in a feed, out of stock in schema and available on the page.

The question is not whether a feed is better than schema. They do different jobs and should describe the same item.

What each source does

A Merchant Center feed is a structured catalogue submitted to Google. It carries identifiers, titles, descriptions, prices, availability, images, links and variant attributes across the inventory.

Product structured data sits on the product page. It describes the item and offer attached to that URL. Google says product data can be provided through structured data, a Merchant Center feed, or both. It also notes that using both maximises eligibility for shopping experiences.

The visible page remains the customer-facing source. Feed and markup should not become a private version of the offer that the buyer cannot verify.

Audit identity before copy

Start with identifiers: id, SKU, GTIN, MPN, brand and variant grouping. If two systems cannot agree which item they describe, polishing a title is premature.

Then compare price, currency, availability, condition and URL. These values change often and produce the most expensive mismatches. Check sale periods and variant URLs separately; a parent product price compared with a selected variant price can look wrong while both systems behave as configured.

Only then compare titles and descriptions. They need not be character-for-character duplicates, but they must refer to the same product and variant. Google’s landing-page guidance explicitly allows some differences while requiring consistency of the underlying offer.

A practical mismatch report

Export the feed and crawl a representative set of live product URLs. Create one row per sellable variant with columns for feed value, page value, schema value and status.

Use four statuses:

• exact match;
• acceptable presentation difference;
• stale value;
• different product or variant.

Do not collapse everything into “pass” or “fail.” A shortened page title is not equivalent to the wrong availability.

Prioritise mismatches affecting identity, purchase conditions or high-revenue products. Fix the system that originates the bad value rather than patching the export. A nightly feed transformation can hide a broken catalogue for months.

What this can and cannot improve

Consistent product data gives machines clearer, current inputs and can improve eligibility. It also reduces disapprovals and prevents customers seeing an offer that changes after a click.

It does not guarantee a rich result, ranking or AI recommendation. Google’s structured-data policies are direct about eligibility not being a promise of display. Treat clean product data as infrastructure: necessary in many shopping journeys, sufficient in none.

Use the structured data checker on a live product page, then compare its values with the exported feed rather than reviewing either source alone.

Sources

• Google Search Central, Product structured data.
• Google Merchant Center, Supported structured data attributes.
• Google Merchant Center, Landing page requirements.