Building a Google Shopping feed that actually performs is mostly unglamorous work. Rewriting product titles so they match how people search. Filling in missing GTINs, brands, and size attributes. Mapping hundreds of products to the right Google taxonomy category. Auditing a 3,000-row spreadsheet for formatting errors before uploading.
None of this is technically complex. All of it is time-consuming when done manually, and error-prone when rushed.
Claude can do most of it faster than you can, and more consistently. Here is how to use it across each part of the feed-building process.
Writing Product Titles That Match How People Search
The product title is your primary relevance signal for Shopping. Google uses it to match your products to search queries. A title like “Blue Jacket Men’s L” will match fewer searches — and less relevant ones — than “Men’s Waterproof Hiking Jacket - Navy Blue, Size Large, Hooded.”
The formula for a strong Shopping title is: Brand + Product Type + Key Attributes (color, size, material, gender, quantity). The order matters — the most important terms go first because Google gives more weight to the beginning of the title.
Rewriting titles manually across hundreds of products is where most people cut corners. Claude handles this at scale.
Prompt example:
I am building a Google Shopping feed for a skincare brand. Below are the product names from our Shopify store. Rewrite each one as a Google Shopping-optimized title following this format: Brand + Product Type + Key Attributes. Keep titles under 150 characters. Do not add attributes that are not implied by the original name.
- Hydra Boost Serum 30ml
- SPF 50 Daily Moisturiser
- Retinol Night Cream 50ml
Claude will return structured, search-optimized titles you can paste directly into your feed. For a catalog of 500 products, paste them in batches and Claude will process each one consistently.
What makes this better than doing it manually: Claude applies the same logic to every product without getting tired or inconsistent. The 400th title gets the same attention as the 4th.
Writing Product Descriptions for Feed Quality
Feed descriptions are less critical than titles for Shopping relevance, but they contribute to match quality and are required for Merchant Center compliance. If your descriptions are copied directly from your website and contain promotional language (“Buy now!”), HTML tags, or duplicate the title exactly, they will create quality issues.
Prompt example:
Write a clean product description for a Google Shopping feed. The description should be factual, 500-700 characters, no promotional language, no HTML, no calls to action. Focus on what the product is and its key features.
Product: Retinol Night Cream 50ml Key features: 0.3% retinol concentration, fragrance-free, suitable for sensitive skin, cruelty-free, dermatologist tested
Claude returns a feed-ready description. Use the same prompt structure across your catalog to keep descriptions consistent in tone and format.
Mapping Products to Google Product Categories
The google_product_category attribute uses Google’s own taxonomy — a hierarchical category system with thousands of entries. Every product should be mapped to the most specific applicable category. Getting this right helps Google understand your products and can affect which required attributes apply to your listings.
Doing this manually means looking up categories in Google’s taxonomy browser for each product type. For a multi-category store with hundreds of product types, this is hours of work.
Prompt example:
Map each of the following product types to the most specific applicable Google product category from Google’s Shopping taxonomy. Return the full category path and the numeric category ID.
- Retinol night cream
- Waterproof hiking jacket
- Stainless steel water bottle
- Men’s leather wallet
- Wireless noise-cancelling headphones
Claude knows Google’s product taxonomy and will return the correct category path and ID for each product type. Verify a sample against the official Google taxonomy, then apply the IDs to your feed.
Generating a Feed Template From Scratch
If you are building a feed for a new store or a new product category and you are not sure which columns to include, Claude can generate the correct structure for you.
Prompt example:
Generate a Google Shopping feed template in tab-delimited format for a women’s clothing store. Include all required attributes and the most important recommended attributes for apparel. Add one example row showing the correct format and accepted values for each column.
Claude returns a feed template with the correct column headers in the right format, required attributes flagged, and an example row showing accepted values. Use this as your starting structure and fill in real product data.
This is particularly useful when setting up a feed for a new product category — apparel has different required attributes (gender, age_group, size) than electronics or home goods, and getting the columns right from the start prevents a wave of disapprovals after the first upload.
Auditing an Existing Feed for Errors
If you have an existing feed with errors in Merchant Center, Claude can help you identify and fix them before re-uploading.
Paste a sample of rows from your feed (or the error list from Merchant Center Diagnostics) and ask Claude to identify issues:
Prompt example:
Here is a sample from my Google Shopping feed. Identify any formatting errors, missing required attributes, or values that do not match Google’s accepted format. Flag each issue with the column name, the affected value, and the correct format.
[paste 5-10 rows from your feed]
Claude will flag things like availability values that are not in lowercase underscore format, prices missing a currency code, image URLs that appear to contain query parameters that might cause fetch issues, and attribute values that do not match the accepted list.
For large feeds, do this with a representative sample rather than the full file. If Claude flags a formatting issue in 5 sample rows, the same issue likely exists across the full catalog and you can fix it at the source.
Bulk-Filling Missing Attributes
One of the most common feed quality issues is missing attributes that are required or strongly recommended for your product category. GTINs, brands, sizes, colors, materials — these frequently exist in a spreadsheet somewhere but have not been added to the feed.
If you have product data scattered across multiple sources (a product database, supplier sheets, your CMS), Claude can help consolidate and format it.
Prompt example:
I have a list of products with incomplete feed data. For each product, I will give you what I know. Add the missing attributes based on the product information, following Google Shopping feed requirements. Flag any attributes you cannot determine from the information provided.
Product: Nike Air Max 270 Men’s Sneaker, Black/White, Size 10 US Known: id=NKE-AM270-BW-10, price=130.00 USD, availability=in_stock, link=[url], image_link=[url] Missing: brand, gtin, google_product_category, gender, age_group, color, size, condition
Claude fills in what it can confidently determine (brand=Nike, gender=male, age_group=adult, color=Black/White, size=10, condition=new, google_product_category=3 or the appropriate footwear path) and flags the GTIN as something you need to source from the product packaging or supplier.
This is most useful for private label products or products from suppliers who do not provide structured data — Claude can generate the attribute values you would otherwise need to research and enter manually.
Writing Custom Labels for Campaign Segmentation
Custom labels (custom_label_0 through custom_label_4) are free-form attributes in your feed that you define yourself. They do not affect how Google categorizes your products — they are used in Google Ads to create campaign subdivisions. Common uses: margin tier, product category for internal segmentation, sale status, season.
If you need to assign custom labels across a large product catalog based on rules (all products under $20 get label “low_price,” all products with margin over 40% get label “high_margin”), Claude can help write the logic.
Prompt example:
I want to add custom_label_0 to my Google Shopping feed to indicate margin tier. Rules: products with margin above 50% get “high_margin”, 30-50% get “mid_margin”, below 30% get “low_margin”. Write a Google Sheets formula I can apply to my feed spreadsheet to populate this column based on the margin percentage in column G.
Claude returns a formula you can apply directly in Google Sheets. The same approach works for any rule-based labeling — seasonal tags, clearance flags, bestseller labels.
What Claude Cannot Do for Your Feed
A few honest limitations:
Claude does not have access to your actual product data unless you paste it into the conversation. It cannot pull from your Shopify store, read your Merchant Center account, or fetch a live feed file. You need to provide the raw data.
Claude cannot verify GTINs. It can tell you that a GTIN field should be a 12 or 13-digit number in a specific format, but it cannot confirm whether a specific GTIN is correct for a specific product. Verify GTINs against the product packaging or your supplier’s data.
For very large catalogs — tens of thousands of SKUs — you will hit practical limits on what you can paste into a single conversation. Work in batches, or use Claude’s API to process feed data programmatically at scale.
The Practical Workflow
For a new feed build:
- Export your product catalog from your CMS or backend in whatever format it is in
- Ask Claude to generate the correct feed template structure for your product category
- Paste your raw product data and ask Claude to map it to the correct feed columns with the right formatting
- Ask Claude to rewrite product titles following the optimized format
- Have Claude generate descriptions for any products with missing or non-compliant descriptions
- Map products to Google product categories using Claude in batches
- Upload to Merchant Center, check Diagnostics, paste the error list back to Claude to diagnose remaining issues
For an existing feed with errors:
- Export the Merchant Center Diagnostics error report
- Paste the error list to Claude and ask it to explain each error type and the correct fix
- For bulk fixes, ask Claude to write the corrected values or Google Sheets formulas to fix them at scale
- Re-upload and verify
The feed work that used to take a full day — template setup, title rewrites, category mapping, description writing — now takes a few hours. The quality is more consistent because Claude applies the same rules to every product without the drift that comes from doing the same task for hours manually.
Feed quality is where Shopping performance is actually won. This is how you get it right without it consuming your entire week.
Related Posts
The Most Common Feed Errors in Google Merchant Center (and How to Fix Them)
How to Use Merchant Center Diagnostics to Audit Your Product Feed
Product Feed 101: What It Is, How It Works, and Why It Matters
Need Help With Your Google Ads?
I help e-commerce brands scale profitably with data-driven PPC strategies.
Get In Touch