Most google shopping ads optimization advice goes straight to bids and budgets. The feed barely gets mentioned, yet it decides which auctions your products enter before a single bid is placed. This is the field-by-field work that moves ROAS, in the order it pays off.
The feed is the real keyword list in Shopping campaigns
Shopping campaigns have no keywords. Google decides which queries your products match by reading the feed: title, description, product type, Google category, and attributes. Whatever lives in those fields is your targeting, whether you wrote it deliberately or let an app auto-generate it.
This is why two stores selling the same product at the same price can get completely different Shopping results. One feed says "ACME-2041-BLK" and matches almost nothing relevant. The other says "Black Leather Crossbody Bag for Women" and enters every auction that matters.
In the accounts I audit, the feed sits behind a large share of query mismatch: products showing for searches they will never convert on, and missing from searches they would win. No bid strategy can fix entering the wrong auctions. The product feed FAQ has the short version of why.
Title engineering: the formula that gets products into the right auctions
The title is the heaviest-weighted text field in the feed, and the first 70 characters carry the most influence because that is what shows in most placements.
The structure that performs across my accounts: lead with what the product is and the attributes buyers search by. For apparel that usually means brand, gender, product type, color, material. For hard goods: product type, key spec, size, brand. The pattern shifts by category, but the principle holds everywhere. Front-load the words a buyer would type.
What kills titles: starting with a brand name nobody searches, stuffing every keyword variant into one line, or copying the website's stylized product name. "The Wanderer" tells Google nothing. Rewrite the titles on your twenty highest-traffic products first and you will see the auction mix shift within weeks.
The attributes Google weights most (and the ones everyone forgets)
After the title, four fields do most of the work:
- Product type. Your own taxonomy, so be specific. "Home > Bathroom > Walk-In Showers" beats "Showers". Google reads it as a matching signal and you will use it later for campaign segmentation.
- Google product category. Pick the deepest accurate node, not the parent category that was quickest to assign.
- Description. Read by the algorithm even though shoppers rarely see it. Write the use cases and search-relevant detail the title cannot fit.
- Color, size, material, gender. The forgotten ones. They power filtered and refined placements, and missing values quietly remove you from those auctions entirely.
Completing these for an entire catalogue sounds tedious because it is. It is also the cheapest reach you will ever buy.
GTIN coverage and category depth: eligibility left on the table
GTINs match your products to Google's product graph. With a valid GTIN, your offer joins the aggregated listing with review data and price context, and it becomes eligible for placements that unidentified products never see.
Stores skip GTINs because Shopify does not force the field. If your products have manufacturer barcodes, add them to the barcode field and map them to the feed. If you manufacture your own goods without GTINs, set identifier_exists correctly instead of leaving the field blank and collecting warnings.
Category depth compounds the same way: deeper, accurate categories enter more specific auctions where competition is thinner and intent is higher. Both fixes are one-time work that keeps paying.
Merchant Center approval is not google shopping ads optimization
A green checkmark in Merchant Center means your feed meets policy minimums. It says nothing about whether the feed is competitive.
An approved product with a vague title, no GTIN, and a shallow category will run. It will just run in the wrong auctions at a worse cost. Treating approval as the finish line is the most common feed mistake I see in audits.
The benchmark is not "approved". It is "does this product enter the exact auctions where my best customers search". On a glamping brand account, feed rework came before the jump: revenue grew 283% after the catalogue was rebuilt to match how buyers search instead of how the warehouse named products.
Feed errors that silently suppress impression share
Beyond outright disapprovals, a set of quieter issues caps your reach:
- Price or availability mismatches between feed and site, which trigger item-level throttling long before any policy email arrives.
- Image issues: promotional overlays, watermarks, or placeholder images on variants.
- Stale feed syncs, where a sale ends on the site but lives on in the feed for days.
- Duplicate products left over from app migrations, splitting performance history across two offers.
Each one trims impression share a few percent at a time. Stacked together, they explain why an account can look healthy in diagnostics and still underdeliver against competitors with cleaner pipes.
Prioritizing feed work when you have 500+ SKUs
You will not hand-rewrite a thousand titles, and you do not need to. Sort by revenue and fix in tiers.
Tier one: the top products driving most of your Shopping revenue. Hand-write those titles, complete every attribute, verify GTINs one by one.
Tier two: the mid-tail. Fix with rules in Merchant Center or a feed app, using templated title structures per product type so hundreds of SKUs improve in one pass.
Tier three: the long tail. Correct the categories, let the templates handle the rest, and move on.
This is standing work, not a one-off project. New products launch with weak data, suppliers change barcodes, apps overwrite fields. It is part of what ongoing management is for, the same way query review is.
Fix the input before tuning the machine
Bids, budgets, and targets tune how aggressively you compete in auctions. The feed decides which auctions you are in. Polishing the first while ignoring the second is tuning an engine that is pointed the wrong way.
Run the title and attribute pass on your top sellers, then let the performance data tell you what to fix next. My Google Ads management includes feed work as standard, and the free 48-hour audit will show you where your feed is costing you right now.