A product feed says a jacket costs $89 and is in stock. The product page shows $99 and “sold out.” Its JSON-LD still says $89 and InStock. Which value should a shopping system trust?
This is a concrete data problem, not a mysterious AI visibility score. Google Merchant Center compares product data with landing pages and structured data; price or availability differences can trigger warnings, automatic updates, or disapproval, depending on the case and settings. Other shopping systems have their own data paths. A clean Google feed alone does not prove placement in ChatGPT or any AI assistant.
Check one actual product, end to end
Choose a product with a stable URL, active offer, and variants if relevant. Record the time, country, currency, product ID, URL, and selected variant. Then compare four views:
| Surface | What to record |
|---|---|
| Product page | Visible title, selected variant, price, currency, stock, shipping promise |
| Page markup | Product/Offer name, SKU or GTIN where applicable, price, priceCurrency, availability, URL |
| Product feed | Item ID, link, price, currency, availability, variant identifiers |
| Merchant Center diagnostics | Item status, mismatch reason, last crawl or update timestamp |
Use the Google Rich Results Test for the page markup and compare its extracted values with what a customer sees. Then inspect the corresponding feed item rather than assuming the feed generator is correct. A logged-in price, geolocation rule, currency switcher, or cache layer can change what different visitors receive. Test the same destination country and variant that the feed item represents.
Common causes worth checking
Stale feed schedule. The store changed inventory or price after its last feed export. Shorten the update interval or push item updates when the source of truth changes.
Variant mismatch. The feed represents a blue size M jacket, while the URL or default page selection displays black size L. Make variant links and attributes explicit, and verify the selected offer on landing.
Cached markup. The visible page updated, but JSON-LD was rendered from an older cache entry. Ensure price and stock on the page and markup use the same current source and cache invalidation.
Currency or market mismatch. The feed targets one country and currency, but the landing page redirects or defaults to another. Check market-specific URLs, shipping settings, and user-agent behavior.
Sale-price confusion. A struck-through price, member-only price, coupon, or time-limited sale can be represented differently across feed and page. Compare the price a buyer can actually obtain under the feed's conditions.
Fix the underlying catalog and synchronization path first. Google's automatic item updates may correct some differences, but they are a fallback, not a substitute for accurate source data; Google says a product may still be disapproved instead of updated.
Verify the repair
Rebuild or resync the affected item, clear relevant caches, and check the page, markup, and feed again at the same timestamp and market. Review Merchant Center diagnostics after its next processing cycle. Repeat with one product from each major category and with variant families, sale items, and out-of-stock items. Keep a dated before-and-after list of exact URLs and item IDs so a future recheck compares the same products.
If you want a fast starting point, use Lumen's free product-data scorecard, then run its free three-product public scan. The scorecard is a self-reported worksheet, and the scan is a limited public sample. Lumen's optional $299 one-store assessment goes deeper with a representative sample of up to 100 products and a documented improvement plan. Neither guarantees a particular ranking, AI appearance, traffic level, or sale.