Product image, price, and store name pulled directly from your Merchant Center feed. The base layer of e-commerce paid search — and the format where feed quality matters more than bid tactics.
Shopping ads show on the Google results page as product cards with an image, title, price, and store name. They're triggered by product-related queries — "running shoes size 42," "ceramic coffee dripper," "ergonomic office chair under 300 euro."
The ads themselves aren't written by us, or by you. They're generated automatically from the Merchant Center product feed. That makes the feed the single most important asset in any Shopping setup, and the place we spend the most time before ever touching a bid.
Users see image and price before clicking, so by the time they land on the product page they're a much warmer visitor than a Search-Ads click.
A 2,000-SKU catalogue is automatically searchable across thousands of long-tail queries without anyone having to write a keyword for each one.
The most specific queries — colour, size, model number — are exactly where Shopping outperforms generic Search.
A properly configured Merchant Center also gets you free organic Shopping listings — separate inventory, no media cost. Easy upside.
Most underperforming Shopping accounts have feed problems hiding in plain sight — disapprovals, missing GTINs, weak titles, wrong product categories. We start with a feed audit and fix what's there before changing anything in the campaigns.
The product title is the single most important field in the feed. We restructure titles to lead with the attributes Google's matching algorithm actually weights — brand, type, size, colour, model.
Splitting the catalogue into too many campaigns starves each of conversion data. Splitting it into too few hides important performance differences. We segment by margin band, brand, or category — whichever maps to how you actually think about the business.
Both campaign types use the same feed but behave differently. We use Standard Shopping when bid control and surface transparency matter, and Performance Max when the goal is automated scale across surfaces. Often the right answer is to run both, on different parts of the catalogue.