Product data and Merchant Center
Google Merchant Center and product feed optimisation.
Fix eligibility and store-data mismatches first. Then improve stable product IDs, variants, titles, taxonomy and other accurate attributes that help the right product match relevant demand. Review the results at product level.
Policy repairs and commercial experiments have different review conditions.
Merchant Center workflow
Fix eligibility, clarify products, then review the evidence.
Follow an order of work.
Identify affected inventory
Separate account-wide restrictions from individual product issues. Group by reason and product family, then prioritise relevant in-stock inventory with defensible economics.
Reconcile the same variant
Compare the storefront, platform export and Merchant Center record. Record where each field comes from; recurring mismatches may require a source-system repair.
Test a defined improvement
Keep dated exports and a consistent product set. Record stock, promotion and campaign changes so the result can be interpreted.
Keep product identity stable.
- Keep product IDs stable through routine updates; do not change them simply to clear an issue.
- Use genuine GTIN, MPN and brand values where applicable. Do not invent identifiers.
- Group variants consistently and align size, colour and other distinguishing attributes with the selected landing-page variant.
- Represent price, currency, availability and delivery accurately across the feed, store and checkout.
Write titles that describe the actual item.
Put the relevant item type and distinguishing attributes in an order buyers can interpret. A furniture item may need dimensions and material; fashion may need garment type, material, colour and size; pet food may need species, life stage and pack size.
For example, “Oak dining table, 180 cm, natural finish” is only appropriate when every attribute is true. That is a factual wording example, not an experiment result. Do not append a list of popular queries. Investigate ambiguous wording when it attracts irrelevant demand.
Give each product-data field a job.
Taxonomy
Use product_type for useful store taxonomy and the applicable Google product category specification for classification. They are different fields.
Images and destinations
Show the exact advertised product and make sure the selected variant and destination match.
Custom labels
Use labels for decisions you can act on, such as contribution bands, seasonality, stock depth or controlled test groups.
Cost-of-goods data
Check coverage and meaning where COGS is used. It is not total variable cost and does not turn revenue ROAS into net profit.
Close the product-level feedback loop.
Maintain a small decision log: product group, issue, evidence, proposed change, owner, date and review condition. Keep other promotions, stock changes and campaign changes visible.
Review relevant search demand, product performance and store conversion together. More traffic without more useful orders can indicate price or product-page uncertainty rather than another feed wording problem.
Stop or reset an experiment when implementation is incorrect, the product set becomes incomparable or stock changes invalidate the decision. A promotion-driven revenue increase alone does not prove the feed change worked.
Continue the acquisition review.
Official documentation.
Use the current platform guidance when planning implementation changes.
Need to decide what to fix first?
The €650 eCommerce Diagnostic reviews the wider account and turns the evidence into a prioritised roadmap and review call. Ongoing management is scoped around catalogue, markets and implementation needs.
Written by Andrei Visan.
I’ve worked hands-on with Google Ads since 2010. This resource explains a diagnostic approach, not a claim of specific client results.
Find the next useful account action.
Bring the store and the acquisition problem. We can confirm whether the €650 diagnostic or another next step fits.