Optimizing feed titles and attributes for Shopping ads

Here's the fact that reframes Shopping campaigns: there are no keywords. Google decides which queries your products enter based on your feed content - above all, the title. Which means feed optimization is targeting, and the difference between a mediocre and a strong feed shows up directly as impression volume, relevance, and cost. This is the craft lesson.

Titles: the highest-leverage 150 characters in paid

Shopping titles work differently from your on-site titles: they're matched against queries, truncated around 70 visible characters (though 150 are indexed), and weighted front-to-back. The working pattern for apparel: brand + product type + key attributes (material, color, gender/cut) - "Yourbrand Linen Shirt Relaxed Fit Men Navy" - front-loading whatever your customers query most. Compare that against what stores commonly send: the on-site title, verbatim, mood-name and all ("Anine - Navy"). The site title has a job (Phase 4); the feed title has a different job, and this divergence is the single strongest argument for the feed-tool transformation layer from Phase 3: build Shopping titles by rule from structured fields (brand + type + material + color from your metafields), and every product gets an optimized title automatically - no copywriting, just projection. Yes: this only works if the attributes exist. The Phase 2 rent, collected again.

The attributes that decide auctions

Beyond the title, the fields that most affect performance: google_product_category (be precise - "Apparel & Accessories > Clothing > Shirts" beats the parent category; precision improves query matching), product_type (your own taxonomy string - used for both matching signals and, critically, campaign segmentation later), color/size/gender/age_group (mandatory for apparel; also power the filter facets in Shopping's UI - an item without color never appears to anyone filtering by color, the feed version of Phase 5's facet lesson), GTIN (unlocks matching to Google's product graph - better query mapping, eligibility for popularity signals), and sale_price as a distinct field (activates the strikethrough price display, a measurable CTR lever).

Test like it's targeting, because it is

Feed optimization earns iteration the way ad copy does. The practical loop: change one thing per cohort (a title structure, say), let it run two-plus weeks, and read the Shopping query reports - did impression volume rise, did query relevance improve (the queries you're matched to are visible in the search terms report), did CTR move? A/B discipline is hard in feeds (no clean control groups), but before/after on stable cohorts is honest enough to steer by. And the meta-habit: the search terms report is also demand data - queries converting in Shopping are validated targets for your organic collection work. Phase 4's loop, closing again from the paid side.