Quick answer: Audit a Merchant Center feed by comparing the submitted product record with the selected variant, visible landing page and purchasable checkout offer. Fix identity, price, availability, images and delivery inconsistencies at their source. Validate a representative sample before changing the whole catalogue, then monitor the account’s actual issue details and product eligibility.
Better product discovery starts with a reliable offer. A store can publish persuasive product descriptions while submitting the wrong variant, an outdated price or a link that opens a different item. These errors become more consequential when shopping interfaces assemble information from several sources. A practical feed audit follows the product from its source record to a real buying decision.
Official product documentation checked on 21 October 2026. Product examples below are fictional.
What should a product feed audit establish?
Google’s product data specification describes the attributes used in Merchant Center and explains that inaccurate, conflicting or missing information can affect product eligibility and presentation. Requirements vary by product, destination and market. Check the relevant specification rather than applying a generic list without context.
The audit should establish one clear source for each important fact, a consistent representation across systems and a repeatable way to detect errors. It should not simply produce a longer spreadsheet of copied fields. Every recorded discrepancy needs a cause, an owner and a method of verifying the correction.
Choose a sample that exposes different failure modes
Include a standard product, a variant family, a sale item, an out-of-stock item and a product with special delivery conditions. Add a recently edited item and a record with a known account warning. This sample is more informative than checking only the easiest best-seller with a single size and a stable price.
Preserve the original data export before editing. Record the product identifier, exact submitted URL, market and observation time for each sample. When prices or stock change frequently, time matters. Two systems can appear inconsistent because the comparison was made across different update moments rather than because either field was permanently wrong.
| Field group | Compare these sources | Typical investigation |
|---|---|---|
| Identity | Catalogue, feed and selected variant | A size or colour maps to the wrong record |
| Offer | Feed, page and checkout | A sale ended in only one system |
| Availability | Inventory, feed and purchase control | A sold-out item remains purchasable |
| Images | Submitted image and visible variant | The picture shows a different configuration |
| Delivery | Submitted settings and actual destination | A region cannot receive the advertised service |
Keep product identity stable and specific
Check whether an identifier consistently refers to the same item. Reusing one identifier for unrelated products makes history and diagnostics harder to interpret. Preserve identifiers as text during spreadsheet work, especially when they contain leading zeros or long digit sequences. Confirm that exporting and reopening the file does not silently change them.
Use genuine manufacturer identifiers where applicable and verify requirements for the product category. Do not invent a GTIN to fill an empty cell or copy the identifier from a similar-looking item. If the product lacks a particular identifier, follow the relevant documented rules and record the actual product circumstances.
Check variant selection from the submitted URL
Open the exact feed URL in a clean browser session. Confirm that the intended size, colour or other variant is selected and that the offer matches that selection. Test a direct visit, not only navigation from a parent product page that may carry state in the browser. A reviewer should be able to reproduce the same product selection from the submitted address.
Google’s product variant guidance explains how ProductGroup and related properties can describe variants for supported search experiences. Structured data and feed configuration are connected parts of an offer, but one should not be assumed to repair mistakes in the other. Verify the representation your actual storefront outputs.
Reconcile price and promotions through checkout
Compare the submitted amount, visible price, selected quantity and final purchasable offer. Check the currency and any conditions attached to a sale or promotion. A discount shown only to logged-in members should not be described as an unconditional offer to every visitor. Record the exact eligibility and time window involved.
Investigate the source of recurring mismatches. A feed scheduled before a price update, a cached page and a separate promotional plugin can create several versions of the same offer. Fixing one exported cell may be overwritten on the next synchronization. Correct the system or timing that generated the disagreement.
Follow stock changes from inventory to the buyer
Select an item whose availability has changed recently and inspect the update path. Identify which system owns inventory, how the feed receives the change and when the storefront updates. Then test what a visitor can actually buy. An unavailable label and an active purchase button can create confusion even before any shopping platform evaluates the page.
Check variants independently. A product family may have one available size and several unavailable ones. A broad family-level statement can hide the difference. Keep the audit focused on the exact offer being submitted, and make the customer-facing response clear when a selected variant is no longer available.
Make titles and descriptions identify the real product
Use the information a buyer needs to distinguish the item: brand where relevant, product type, important specifications and the specific variation. Remove unsupported superlatives that occupy space without identifying the offer. A title should not imply a feature, material or quantity that the purchased item does not include.
In descriptions, answer practical questions about fit, compatibility, included parts and important limitations. Match those facts to the visible page. Our ecommerce SEO guide explains how product and category information work together, while our image SEO guide covers the surrounding presentation of useful product imagery.
Validate structured data against the visible offer
Use the merchant listing structured-data documentation to review supported Product and Offer information. Test representative pages with the available validation tools, then compare the extracted values with what a buyer sees. A technically valid object can still describe the wrong product or an obsolete offer.
Watch for several plugins emitting competing product objects or inconsistent prices. Identify the system that should own the markup and configure it deliberately. Do not add fictional review counts or ratings to make a validation panel look complete. Include accurate, applicable information that the business can substantiate.
Investigate account issues with a reproducible record
Open the actual issue details in Merchant Center and record the affected items, destination and stated reason. Different issues need different responses. A missing attribute, an unreachable image and a policy concern should not be placed in one unexplained “feed error” category. Preserve the wording needed for the investigation.
After correcting the cause, submit or refresh data through the supported route and follow the available review process where relevant. Check the resulting account state instead of assuming that uploading a new file immediately resolves every issue. Do not guarantee a review time, approval or display position that the store cannot control.
Work through one catalogue mismatch
Consider an illustrative store selling a desk lamp in white and black. The black lamp is discounted, but the feed uses the parent URL, which defaults to white. The submitted image shows black, the landing page opens white, and the checkout price follows whichever colour the visitor manually selects.
The audit should record the mismatch, confirm the intended variant address and align the record, image and offer. Then repeat the direct-URL test in a clean session. The fix is more specific than rewriting the title. It also gives the merchant a reusable check for other variant families before the next promotion.
Make feed quality part of normal store maintenance
Prioritize corrections using the actual scope and customer impact. An error affecting a frequently ordered product family deserves a different response from a cosmetic issue on an unavailable item. Keep the original account severity and affected destinations in view rather than inventing a single score that hides the underlying problem. Assign a specific fix and verify it against the same item and market.
After a source-system migration, compare a small set of known records before accepting the entire export. Check identifier preservation, variant relationships, currency, special characters and destination URLs. A file can import successfully while still changing the meaning of an identifier or stripping a useful product detail. The final review should inspect actual records and pages, not only the transfer’s success message.
Assign responsibilities across catalogue management, merchandising, development and fulfilment. A feed owner needs a clear way to escalate a stock issue or a wrong promotion. Keep a short review around major sales, theme changes and source-system migrations, when several systems may update on different schedules.
Use your store SEO workflow and technical audit process to check the surrounding website. Accurate product data supports clearer discovery and fewer preventable errors. It remains one part of a functioning store, alongside a trustworthy offer, usable pages and reliable order fulfilment.
Common questions about product feed audits
Can structured data replace the need to inspect feed records?
Do not assume so. Review the requirements and data sources for the intended experience, then compare each representation with the actual offer.
Should I invent a product identifier when one is missing?
No. Verify the manufacturer’s information and the category’s requirements. Follow the documented treatment for products without that identifier.
Does fixing a feed issue guarantee free-listing traffic?
No. Correct data can improve eligibility and accuracy, but display, demand and customer behaviour still affect results.
How many products should a small merchant check first?
Start with a representative sample that includes variants, promotions, stock changes and known issues. Expand according to the patterns found.
Join the discussion