Run a repeatable AI visibility audit
Trace sample SKUs from store to feed, identify the cause of stale product facts, and track fixes.
Sample products that stress the catalog
Select a best seller, a newer SKU, an out-of-stock variant, a sale product and a listing with a market-specific restriction. Record their IDs, titles, canonical URLs and variant URLs. Include the date and time for each spot check so you can compare values observed before and after a feed refresh.
Start with three sources of evidence
Compare storefront and checkout, structured data, and any feed accepted by the shopping surface you use. Use the SKU plus variant URL to join your observations. Compare price, currency, availability and relevant policy. Add a screenshot or a snippet from each surface to the finding so the catalog team can reproduce it.
Trace a single mismatch to its owner
Suppose a variant is sold out in checkout and available in the feed. Check whether the feed record has the variant ID, when inventory changed, and which run last exported it. Correct the mapping or refresh schedule and verify the same record again. If the page's structured data is stale too, check page caching and the Theme or plugin that writes the offer.
Measure answer quality using a fixed set of questions
Try prompts asking about features, delivery or comparisons that a real buyer would ask. Include questions where your product is a poor fit. Record the surface, locale, citation URL, product facts and date. Search results vary over time, so use the observations to investigate incorrect factual claims and broken citations.
Prioritize by impact on a purchase
Fix inconsistent price and availability first, then invalid URLs and misleading policy. Add missing dimensions or compatibility when those issues recur. Assign each finding an owner and a retest date. At the next check, use the same SKUs and add one recently changed product to catch regressions.