Retail search advertising is generated from a product data file rather than from keywords. When performance drops without an obvious cause, the file is the usual explanation.
The feed is the campaign
Each row in the file describes one product, supplying its title, description, price, availability, images and identifiers.
The advertising system builds the ad from those fields and decides which searches the product may appear against, so the file performs the role keywords play elsewhere.
An item missing from the file, or rejected by it, is simply absent from the market with no campaign-level indication that anything changed.
Disapprovals are silent at the account level
Individual items are reviewed and can be rejected for a mismatched price, missing identifier, image problem or policy issue in the description.
The campaign continues serving with the remaining items, so overall spend and returns fall gradually rather than stopping in a way that triggers attention.
Because the reporting shows fewer impressions rather than an error, the drop is frequently misdiagnosed as competitive pressure or seasonal softness.
Price and availability mismatches are the common cause
The price in the feed has to match the price on the landing page, and a store running a promotion that updates the site faster than the feed creates a discrepancy.
Out-of-stock items that remain in the file draw clicks to a page that cannot fulfill them, spending money on traffic that could not convert.
Both problems originate in synchronization frequency rather than in the advertising account, which is why they persist through campaign optimization work.
Titles determine what a product competes for
Query matching relies heavily on the product title, and titles written for a website's own layout often omit the words shoppers actually search.
Brand, product type, and defining attributes such as size or color belong near the front, because matching and truncation both favor early terms.
Rewriting titles alone can change which searches a catalog reaches, without any change to bids, budgets or campaign structure.
Monitoring the file is the preventive measure
The useful checks are the number of items approved versus submitted, the count of disapprovals by reason, and the age of the last successful update.
Alerting on a fall in approved items catches most incidents within a day, which is usually before the effect becomes visible in revenue.
Advertisers who monitor only campaign metrics discover these failures weeks later, by which point the lost sales cannot be recovered.