Google Shopping Feed Optimisation: How Product Data Drives Performance
In Google Search, keywords control which queries trigger your ads. In Google Shopping, the product feed does that job. Feed quality determines query match, impression share, and which products are eligible to show at all. This guide covers the fields that move results most: product titles, supporting attributes, custom labels for bid segmentation, and keeping the feed free of disapprovals.
- In Shopping, the feed does the job that keywords do in Search; feed quality determines query match, not campaign settings.
- Product titles are the highest-impact field; include brand, product type, key attributes, and size or variant in every title.
- Custom labels allow bidding segmentation by margin, bestseller status, and stock level without affecting query matching.
- Disapproved products show zero ads; the Merchant Center Diagnostics tab must be reviewed weekly, not monthly.
Why the feed is the campaign
In a Google Search campaign, a strategist writes keywords to control which queries trigger ads. In Google Shopping, that control lives in the product data. Google reads the feed and decides which queries each product is relevant to, which competitors it will appear alongside, and at what price it will bid in the auction. No amount of campaign structure or bidding sophistication overcomes a weak feed, because the feed is what the entire system reads from.
The implication is that Shopping optimisation work happens substantially outside the Google Ads interface. A well-structured feed with accurate, attribute-rich product titles and clean data fields will outperform a poorly-structured feed with better bids. In most e-commerce audits we run, product titles and attribute fields have gone untouched since the store first synced its feed, while bids and budgets get adjusted weekly.
How Shopping matches queries to products
Google does not use a keyword list to match Shopping ads to queries. Instead, it reads each product’s feed attributes, particularly the title, description, product category, and product type, and uses a relevance model to decide which products to enter into the auction for a given search. This means the words in your product title function like broad match keywords: they expand or constrain which queries can trigger your ad.
A product titled “Blue Running Shoe Men” will match a narrower range of queries than “Nike Air Zoom Pegasus 41 Men’s Running Shoe Blue Size 10.” The second title includes brand, model, gender, activity, colour, and size: the attributes buyers use when they have purchase intent. Including them does not just improve match rate; it improves match quality, because the queries matched are more specific and convert at higher rates.
Product titles: the highest-impact field
A working title structure for most product categories: Brand + Product Name + Key Variant Attributes + Size or Quantity where relevant. The specific order of attributes matters less than their presence, but leading with brand (where the brand is known) allows Google to match branded queries and signals product authenticity.
- Include the attributes buyers actually search for. For clothing: gender, colour, material, size. For electronics: model number, storage, colour. For home goods: material, dimensions, room type. Review your Search Terms report in Shopping to see which attribute combinations are appearing in matched queries.
- Use natural language, not internal codes. An internal SKU or product code in the title does not match buyer language and wastes character space. Google allows up to 150 characters; most high-performing titles use 70 to 120 with no padding.
- Do not keyword-stuff. Repeating the same word multiple times or adding unrelated terms that do not describe the product violates policy and triggers quality penalties. Every word in the title should be a genuine product attribute.
Supporting data fields
| Field | Priority | Common Error |
|---|---|---|
| Google Product Category | High | Using a top-level category (Apparel) instead of the most specific applicable one (Apparel > Clothing > Activewear > Shorts) |
| Product Type | High | Leaving blank; this field feeds Smart Shopping and PMax’s category understanding |
| GTIN / MPN | High for branded products | Missing on branded products that have one; triggers disapprovals and excludes from GTIN-matched auctions where Google confirms product authenticity |
| Images | High | Low resolution, watermarked, or lifestyle-only images without a clean product shot; main image must be the product on a white or neutral background for most categories |
| Condition | Required | Defaulting to ‘new’ for refurbished stock, which violates policy |
| Description | Medium | Copying from the page meta description rather than writing a feed-specific description that includes key search terms naturally |
Custom labels for bidding control
Custom labels give you a mechanism to segment your product catalogue for campaign and bid management without affecting query matching. The most commercially useful segmentations:
- 01
Margin tier
Label products as high, medium, or low margin. Set higher ROAS targets for low-margin products and allow lower ROAS targets for high-margin ones, since the revenue-per-click threshold differs. Without this segmentation, your bidding treats a 5% margin product identically to a 40% margin product.
- 02
Bestseller or top performer
Label the products that convert at above-average rates historically. These warrant their own ad groups with dedicated budgets and bidding separate from the long-tail catalogue.
- 03
Seasonal or promotional status
Label products that are in a sale, seasonal window, or overstock clearance. Apply temporary bid uplifts or lower ROAS targets for the duration, then revert automatically by removing the label.
- 04
Stock level
Label low-stock products to reduce bids and avoid selling out at full advertising cost. Label high-stock items where clearance speed matters more than margin.
Custom labels are applied in the feed and read in the campaign. Changes to labels take effect when Google re-crawls the feed, typically within 24 hours with a scheduled fetch. The campaign structure built on top of them connects directly to the bid strategy work in Target CPA vs Target ROAS. E-commerce Shopping performance is a significant part of our e-commerce work and sits inside our paid media management service.
Feed health and disapprovals
Disapproved products do not show. A feed where 15% of products are disapproved is a campaign running at 85% capacity at best. Merchant Center’s Diagnostics tab shows disapprovals by reason and count; it should be checked weekly, not monthly, because new products are added and rules change. The four disapproval reasons that affect the most accounts: missing or invalid GTINs, landing page not matching the advertised price, image policy violations, and destination URL returning a 4xx error. Each is a data-field correction, not a rebuild: add the GTIN, match the landing page price to the feed, replace the flagged image, or fix the broken URL.
Frequently Asked Questions
A Google Shopping product feed is a structured data file submitted to Google Merchant Center that contains your product inventory: titles, descriptions, prices, images, categories, GTINs, and other attributes. Google reads this file to understand what you sell, matches it against search queries, and determines whether your products are eligible to appear as Shopping ads.
The product title is the most important field in the feed. It is the primary signal Google uses to match your product to search queries, and it is the main visible text in the Shopping ad. Including the key attributes in a consistent order, brand, product type, key specifications, colour, size where relevant, significantly expands the range of queries the product matches while maintaining relevance.
Custom labels are free-text fields (0 through 4) in the feed that you define and control. They do not affect query matching, but they allow you to segment products for bidding in your campaign. Common uses include labelling by margin (high-margin, low-margin), by season, by stock level, or by bestseller status, then setting different ROAS targets or bid adjustments per segment.
Common disapproval reasons include missing or invalid GTINs for products that require them, image policy violations (text overlays, watermarks, placeholder images), price mismatches between the feed and the landing page, and destination URL errors. Disapprovals suppress the affected products entirely, so monitoring the Merchant Center diagnostics tab weekly is the minimum maintenance cadence.
Ready to fix what’s costing you conversions?
We’ll review your acquisition funnel, show you exactly what’s underperforming, and hand you a clear, prioritised plan, whether or not you choose to work with us.
Get a clear diagnosis of your acquisition performance.
- 30-minute review
- Free, with no obligation
- Clear findings and priorities
- You keep the findings
30 minutes · Free · No obligation · You keep the findings