Shopping ads: feed, structure, AI bidding, and diagnostics
What this domain tests
This domain tests how Google AI turns a Merchant Center feed into Shopping ads, and how campaign structure, product data, bidding, and assets control that automation. Expect questions on feed attributes, product groups, campaign priority, Performance Max asset groups, audience signals, search themes, target ROAS, and Merchant Center diagnostics.
The feed is the ad
The product feed supplies title, description, link, image_link, price, availability, brand, GTIN, condition, and category. Required attributes must be present and accurate; price and availability must match the landing page. Automatic item updates can reconcile price and availability when the feed lags, but they do not fix missing identifiers or policy violations. Disapprovals are usually feed or policy issues, not bidding failures. Strong titles put brand, product type, and key attributes first, because AI matches queries and generates creative from that data.
Structure and control
Standard Shopping campaigns use product groups (brand, category, item ID, condition, custom label) and priority settings. When the same product appears in multiple campaigns, the highest-priority eligible campaign can win the auction. Negative keywords exclude queries. Performance Max uses asset groups, audience signals, search themes, final URL expansion, account-level negative keywords, and brand exclusions. Search themes are signals, not exact-match keywords. Asset groups combine text, images, and optional video; AI assembles combinations, so variety and relevance matter.
Bidding and measurement
Smart Bidding, especially target ROAS, needs conversion tracking with values. Set a target the account can realistically sustain, allow the learning period, and avoid frequent target changes. AI campaigns optimize toward conversion value, so feed quality and conversion values are the real levers.
Worked example
A home-goods retailer runs a high-priority clearance Shopping campaign and a low-priority always-on campaign. One SKU shows in stock in the feed but out of stock on the landing page, and its Performance Max asset group has one image and one headline at 350% tROAS. Reasoning: fix availability in the feed or enable automatic item updates; the high-priority clearance campaign should serve that SKU while the sale runs; add more relevant images, headlines, and descriptions to the asset group; confirm conversion values before judging tROAS performance. Do not raise tROAS to compensate for a feed error.
Traps that cost marks
Treating search themes as keywords; expecting instant feed changes without updates; judging AI campaigns during learning; confusing product-group exclusions with campaign negatives; and blaming AI for disapprovals caused by missing GTINs, mismatched prices, or policy violations.