Your shop shows one price. Google found another.

01The warning may describe yesterday’s facts

You open the product page. The price looks right. Merchant Center still reports a mismatch. Before changing the theme, feed, or markup again, check when Google found the disagreement.

Google’s mismatch guidance says the current price may already match because an automatic update corrected it. The report can still refer to the mismatch found at an earlier timestamp. It also warns that an automatic correction does not make the underlying error harmless. [1]

02Choose five products, not the whole catalogue

This is a proposed diagnostic sample, not a statistically representative audit. Start with the flagged item, a sale item, a variant, a recently changed item, and an unchanged comparison item where those exist. If your catalogue lacks one category, record that instead of inventing a case.

You need access to the product record, page markup, and Merchant Center data, plus someone authorized to approve corrections. If you only edit the website, collect the evidence and hand it to the feed owner. Do not create a second source of truth you cannot maintain.

Swipe or scroll to compare

03Follow the change until the facts split

Do not assume the webpage is correct simply because it is visible. Compare every surface with the approved business record. Check the exact variant and currency before treating two numbers as a conflict.

For sale products, Google specifically asks merchants to check sale-price effective dates and timezone. Its remediation guidance also asks for consistency across product data, the website, and structured data. [1] Our worksheet adds an owner and timestamp so that a correction has a clear route back to its source.

Swipe or scroll to compare

04Automatic updates do not own your catalogue

Google says automatic updates can use landing-page structured data and other extracted product data to update some attributes. But they are intended to correct sporadic problems for a small proportion of products. They do not replace regular product-data updates. Frequent price or availability changes can exceed what the feature handles. [2]

That distinction matters when choosing an implementation. A managed store connector may already be sufficient. A manual check may suit an infrequently changed catalogue. More frequent changes may justify inspecting a more direct update path, but this worksheet does not choose an API or feed vendor for you.

05Let automation collect discrepancies before it changes prices

Begin with a read-only reconciliation report. A rule can compare normalized values and timestamps. AI may help summarize permitted diagnostic notes, but it should not guess which price is commercially correct or silently rewrite a live offer.

• Define which products and fields the pilot may read. Keep customer and order data outside the task.
• Record missing values as unknown. Do not turn an absent crawl result into a confirmed mismatch.
• Require a named approver for price, availability, and sale-date changes.
• Keep original values and the actual correction receipt so a failed change can be investigated.
• Stop when identity, currency, authority, or permissions are unclear.

06Measure the repair loop, not a ranking promise

For one normal update cycle, count products checked, confirmed conflicts, false alarms, unresolved cases, and total preparation, review, correction, and maintenance time. Separate the initial setup from recurring work. Compare with the current manual process on a similar sample.

A useful result is a traceable correction and a clearer update path. It is not proof of more traffic, AI citations, sales, or margin. No merchant audit or automation trial was performed for this guide. Keep, simplify, or stop the pilot based on the actual record.

If the reconciliation creates more work than the current connector, keep the simpler method. The goal is consistent product facts with less avoidable owner effort, not another dashboard to watch.