Fix Google Merchant Center Errors Without Inventing Product Data
Good automation corrects representation while leaving product facts under your control. FeedDoctor separates safe formatting proposals from decisions that require the actual catalog, manufacturer or stock system.
Make a copy and establish a baseline
Upload a Merchant-compatible CSV or a single populated XLSX worksheet. Your original upload remains separate from the working data. Record the initial error count, affected product count and warnings. A warning is not automatically a rejection, and an optional improvement is not a promise of better placement.
For example, a file with eight products can contain more than eight findings because one product may have several problems. Conversely, a missing column is a file-level problem that affects all products. Use the product-based summary and the detailed report together rather than comparing unrelated counts.
Understand what a safe proposal actually changes
| Before | Proposed after | Boundary |
|---|---|---|
| In stock | in_stock | Spelling only; no inventory lookup or stock prediction. |
| 19,95 usd | 19.95 USD | The amount and supplied currency remain explicit. |
| Product title | Product title | Leading and trailing whitespace only. |
| product_title header | title | A recognized attribute mapping, shown for approval. |
Review the row, field, before value, after value and reason. Uncheck any proposal you do not want. Applying selected changes updates the working feed and re-runs the same analysis engine. It does not overwrite your uploaded source file or update your shop.
Do not automate missing facts
An absent GTIN must come from a reliable product source if one was assigned. An absent price needs a real selling amount and currency. A blank description needs accurate product information. FeedDoctor does not create these values. Duplicate product IDs also require a decision about product identity; generating a random suffix could detach a product from its history.
The validator cannot establish whether an item is available, whether a delivery promise is realistic or whether a product is allowed in a particular country. Correct those facts in the system that owns them. A stock label should describe the current business state, not the state that produces the fewest warnings.
Verify the change and keep an audit trail
- Apply a small, reviewed group of changes.
- Check that the expected findings moved to resolved status.
- Inspect the product data tab to compare actual values.
- Download the corrected feed and the diagnostic report.
- Keep unresolved errors in your work list; they are not silently removed from the output.
The CSV and XLSX reports include a suggested solution and an applied flag. JSON also includes the change history. Use the original download as your reference for a comparison in a spreadsheet editor. Undo applied changes restores the original parsed values inside the analysis.
Choose an export that preserves the data
Identifiers and GTINs are written as text in XLSX exports so leading zeroes remain intact. Formula-like CSV values are a separate risk: the tool blocks product CSV export while such values remain and offers a text-cell XLSX instead. It does not remove a suspicious product row without your knowledge. Workbook styles, macros and original formatting metadata are not carried into the new export.
Return to Merchant Center for the final result
After correcting your source data, submit it through the intended data source and check Merchant Center again. FeedDoctor does not submit products, request account reviews or guarantee approval. Pricing consistency, image quality, shipping, tax settings and policy decisions need their own checks. Review the scope guide before treating a zero-error baseline as a complete account audit.
References and next steps
These instructions describe the current FeedDoctor implementation. Check the official requirements for your product, target country and destination before submitting data.