Google Shopping Feed Optimization: The Work That Decides What Performance Max Can Do
Google Shopping feed optimization is the work of making product data match how people search: titles, attributes, identifiers, images and custom labels. Shopping and Performance Max can only show what the feed describes, so a weak feed caps results however well the campaigns on top of it are built.
Why does the feed matter more than the campaign settings?
Google matches a Shopping ad to a search mostly by reading your product data. There are no keywords to bid on in the usual sense. The title, product type, attributes and description are the targeting. A title written the way a warehouse labels stock, something like MNS TRK JKT BLK L, competes badly against one written the way a shopper types: men’s black track jacket, water resistant, size large. Performance Max inherits the same limit, because its Shopping inventory reads from the same Merchant Center feed. Most accounts I review spend their attention on bid strategy and campaign structure while the feed sits unchanged from the day the store was connected. Structure matters, but it only redistributes whatever relevance the feed already provides. Improving the feed raises the ceiling for every campaign that draws on it, all at once, which makes it one of the few changes in Google Ads with no downside.
What should a feed audit check first?
Begin with Merchant Center diagnostics. Disapproved products earn nothing and limited products earn less than they should, so both lists deserve a weekly review rather than a surprise at quarter end. Next, check completeness on the attributes Google leans on for apparel and similar categories: color, size, material, gender, age group, and GTINs wherever a manufacturer barcode exists. Missing values quietly remove products from searches where they belong. Then rewrite titles for the products carrying most of your revenue, leading with the words people search and following with the detail that sets the item apart. Images come after that: clean, well lit, product filling the frame, with no promotional text laid over the top. Last, look at anything the store platform cannot export. Supplemental feeds in Merchant Center add fields your storefront does not hold, without touching the product catalog itself. Product type and Google product category deserve a pass as well, since a jacket filed under a vague category ends up competing in the wrong auctions.
Who should own the feed week to week?
Feed work falls between teams, which is why it goes stale. The ecommerce team owns the catalog, the paid search person owns the campaigns, and nobody owns the translation layer in between. Any scope you sign for Google Ads management services should name who maintains titles, attributes, labels and supplemental feeds, how often disapprovals are reviewed, and what happens when a new collection launches. If no owner is written down, assume the feed will drift. New products arrive with warehouse names, seasonal labels stay on items long after the season ends, and stock labels stop reflecting what is on the shelf. The remedy is a short weekly routine rather than a quarterly project: check diagnostics, update labels for stock and margin changes, and rewrite titles for anything new that climbs into the top sellers.
How do custom labels change what the budget can do?
Custom labels are the part of the feed most brands leave empty, and they are what allows the account to treat economically different products differently. Five label slots are available. A useful setup gives one to margin tier, one to stock depth, one to season or launch status, and one to bestseller versus long tail. Campaigns or asset groups can then be split along those lines, so a high margin product with deep stock can run on a lower return target than a discounted item with three units left. Without labels, every product fights for the same budget under the same target, and the system favors whatever converts cheapest, which is often your lowest margin stock. Labels also let you pull spend from items about to sell out, so money stops buying clicks for products you cannot ship. Keep the label definitions in a shared sheet, so the ecommerce team knows what each value means and updates it when margins or stock levels change.
For one athletic apparel brand I work with, units sold in June 2026 rose 43.4% on the same month a year earlier, from 45,365 to 65,039. Volume like that moves through a catalog, and every product sold through paid search depends on data that describes it accurately.
A feed review is part of the free 30 minute growth audit I run, and disapprovals that nobody knew about are a common first finding.
