Google Shopping Feed Optimization: How I Build Titles From Real Search Demand

Last updated: 7 November 2026

Google Shopping feed optimization is the work of shaping your product data so Google can match each product to the searches that buy it. I have managed paid search since 2013, and in ecommerce accounts the feed is often where the clearest fixes sit. This guide shows how I pick title words from real search demand, which Merchant Center attributes I fill first, and how I clear disapprovals.

Quick Answer: What is Google Shopping feed optimization?

Google Shopping feed optimization means improving the product data you send to Merchant Center so your products show for the right searches. Shopping ads do not use keywords, so titles, descriptions, identifiers and categories do the targeting. I build titles from real search volume, fill the required attributes, keep price and stock accurate, and fix disapprovals first.

Here is the short version of what I check in a feed and what drives each decision.

Why does the product feed decide where Shopping ads show?

The feed decides because Shopping ads have no keywords to bid on. Google’s help page on the search terms report states that Shopping targeting does not use keywords. So the words in your title, description and category are what Google reads when it matches a search to a product.

Performance Max works the same way for its Shopping placements. The MiKi Jewellery case study puts it plainly: Shopping placements in Performance Max run on the product feed, so the feed came first. A weak feed gives the bidding system weak material, however good the budget or bid strategy is.

On the CANA Wine Company account, the working rule was to treat the product feed as the campaign. That rule changes the order of work. Before I touch bids or budgets, I check what the feed tells Google about each product.

What does a weak feed usually look like?

A weak feed describes products the way the store owner thinks about them, not the way shoppers search. Titles copy internal names or supplier codes. Brand or GTIN fields sit empty, and the category is a broad guess.

Stock and price can drift too. A product may show as in stock in the feed while the site says sold out. Each of these problems either narrows the searches a product can match or puts it at risk of disapproval.

How do I choose title keywords from real search demand?

I choose them from data that proves people search those words. My three sources are Google Keyword Planner, Semrush search volume and the search terms report from Shopping and Performance Max. Keyword Planner gives estimates of how many searches a keyword gets each month. The Performance Max version of the report shows the search terms that triggered your ads, with landing pages and ad formats.

I do not let an AI tool decide the words. A language model can write a fluent title, but it guesses at phrasing and may swap the term shoppers use for a synonym that gets little search. When a title change is not backed by volume or search terms data, I do not make it. For me, that is the core of Google Shopping product title optimization.

What does search demand data change in a title?

It often changes the spelling or the word order. The Judaica store case study lists 9,900 searches for tallit and 1,900 for tallis. Both spellings are real demand, so the store needs both in titles, copy and the feed. A rewrite that keeps only one spelling throws part of that demand away.

On CANA Wine Company, titles carry the bottle name, proof, size and vintage, because that is what people type. A collector hunting a rare bourbon searches for the exact bottle. A title built on a category name would never match that search.

How do I turn search terms into title changes?

I follow the same routine on every catalog, whatever the platform.

1. Export the Shopping and Performance Max search terms for a date range long enough to include conversions.
2. Group the terms by product and note words that appear in converting searches but not in the title.
3. Check those words in Keyword Planner or Semrush and keep only the ones with real volume.
4. Rewrite the title with the strongest product term first, then brand and the key attribute.
5. Push the new title through a supplemental data source, leave the primary feed alone, and compare results.

What does Google allow in product titles and descriptions?

Google allows titles of up to 150 characters and asks you to put the most important details first. Its title attribute guide notes that users usually see only the first 70 characters or fewer. It also asks you to leave out promotional text such as price, sale dates, shipping details and your company name.

Capital letters for emphasis, symbols and HTML tags are out too. Variant details such as size or color belong in the title as well as in their own attributes. Details without an attribute of their own, such as flavor, should go in the title.

Descriptions can run up to 5,000 characters. Google’s description attribute guide advises placing key details in the first 160 to 500 characters. Good descriptions cover size, material, features and variants, and they skip links, comparisons and store policies.

One rule matters for anyone using AI. Google asks merchants to send titles written with generative AI in the structured title attribute, marked as algorithmic media. If you send both, only the standard title is used. In short, Google expects AI-written product text to be declared.

This table puts Google’s title and description rules side by side.

Source: Google Merchant Center Help, title attribute and description attribute.

Which Merchant Center attributes should you fill in first?

Start with the attributes Google requires for every product: ID, title, description, link, image link, availability and price. The product data specification also makes brand required for most new products. A product missing these cannot be approved, so they come before any title work.

Next come product identifiers. Google recommends a GTIN for every product with one assigned by the manufacturer, and warns that a missing or incorrect GTIN may limit visibility. If a product has no GTIN and no brand and MPN pair, you set identifier exists to no. Never set it to no just to silence a warning on products that do have codes.

Which extra attributes matter for apparel and jewelry?

Apparel sold in the US needs color, gender and age group, under the same product data specification. Jewelry sits inside Apparel and Accessories, so similar rules apply. That is why the MiKi Jewellery case study talks about apparel-style data requirements.

For that store, the team followed color with up to three values, material, age group, size only where it applied, and the GTIN or brand plus MPN. None of this is glamorous work. It decides whether a product is eligible at all, so it comes before any clever title testing.

How do product type and custom labels help bidding?

Product type is your own category path, and Google uses only the first product type value for bidding and reporting. Custom labels add five more groupings, each with up to 1,000 unique values across the account, per Google’s custom label guide. Shoppers never see custom labels, so you can name groups in plain business terms.

On CANA Wine Company, products were grouped by margin and intent rather than product type. The point is to let budget follow profit, not the size of a category. I use custom labels for margin, season and stock depth for the same reason.

How do you fix Merchant Center disapprovals?

Fix them in order of impact, starting with anything that affects the whole account. Google’s guide to issues in Merchant Center says disapproved products stop showing across Google. Products with warnings keep showing, but their performance may be limited, and setup or policy issues can lead to suspension.

You find product issues in the Needs attention tab, where you can filter them and download a CSV of affected items. Each issue shows its impact and the steps to fix it. After you correct the data or the store, you upload the data again, and a successful review removes the disapproval.

Price and stock mismatches need special care. Google can apply a preemptive item disapproval when price or availability in your data does not match the product page. Turning on automatic updates in the Automations tab reduces that risk. On CANA, availability and price were kept accurate so budget did not chase sold out stock.

This is the order I work through when a feed has a pile of issues.

1. Account setup and policy issues, because they can stop every product at once.
2. Price and availability mismatches, because they cause preemptive disapprovals.
3. Missing or wrong identifiers: GTIN, brand and MPN.
4. Landing page and image problems, such as broken links or placeholder content.
5. Title and description warnings, which I fix while I rebuild titles from search data.

Plan for review time. Google says a product under review typically takes up to 3 to 5 business days for Shopping ads and free listings. Reviews of setup and policy issues can take up to 7 days. I schedule feed fixes so a key product is not offline during a sale.

What happened when I treated the feed as the campaign?

Three ecommerce accounts show how this works in practice. None were feed-only projects, so I separate what the feed did from what else changed.

CANA Wine Company: titles built the way buyers search

CANA Wine Company is a family-owned wine and spirits store in Nashville that ships across the US. We ran Shopping and Performance Max alongside Search, with titles that carry bottle name, proof, size and vintage. Products were grouped by margin and intent, and negative keywords were held at account level.

Over the 120-day window in the CANA Wine Company case study, the account spent $10.2K and recorded $177K in conversion value. That is a 17.35x ROAS at $5.97 per conversion. The feed was one lever among several, including targeting matched to the legal shipping map, so I credit the whole setup.

MiKi Jewellery: a new Shopify store where the feed came first

MiKi Jewellery is an independent Shopify brand in Western Australia with a brand-new Google Ads account. Its product data flows into Merchant Center, so the jewelry attributes came first. Then came five Performance Max campaigns, one per collection or product type.

In the first 47 days, the MiKi Jewellery case study reports 7.81 in conversion value per cost, from A$2.88k in spend. One campaign had all its asset groups limited by policy, but separate campaigns kept that problem away from the rest of the catalog.

A Judaica store: two spellings, both kept

For a Judaica ecommerce store, Customer Match and remarketing are not available because religious products fall into a sensitive category, so the Judaica case study calls the feed the main lever. Both transliterations go into titles, copy and the feed. Across the comparison periods, Google Ads ROAS moved from 4.90 to 17.15 while spend fell 37%, and budget also moved toward named-product demand.

Should you let AI rewrite your product titles?

Not without search demand data behind every change. AI is useful for formatting titles at scale once you know which words belong in them. It is poor at choosing those words, because it cannot see your search terms, your volume data or what converts.

Plenty of feed tools now offer AI title rewrites. The risk is quiet: a rewrite can drop the exact term that drove sales, and you only notice when Shopping clicks fall. I keep the original titles, test changes on a set of products, and compare results before I roll anything out.

A supplemental data source makes that testing safe. Google describes it as a secondary source that adds or updates details, and it cannot add or remove products. I put new titles there, so the primary feed from Shopify or another platform stays as it is.

If you do use AI-written titles, follow Google’s rule and send them in the structured title attribute. Then check every suggestion against Keyword Planner or Semrush volume before it goes live. If the words have no volume, the title goes back to the drawing board.

How often should you review a Shopping feed?

Review it whenever the catalog, prices or search data change, and check the Needs attention tab each time you work in the account. Feeds drift. New products arrive with supplier titles, prices change, and items sell out.

I tie title reviews to my search terms reviews. When the Shopping or Performance Max report shows a converting phrase that the titles do not contain, that product goes on my list. I explain how I read that report in my guide to the Google Ads search terms report.

Feed work also connects to campaign structure. If you are deciding how Shopping, Performance Max and Search should share one catalog, read my notes on running Google Ads for an ecommerce store. My guide to setting up a Performance Max campaign covers the guardrails I add before launch.

FAQs

What is the most important attribute in a Google Shopping feed?

The title carries the most weight for matching, because Shopping targeting does not use keywords. Google reads the title to understand what the product is. Put the product term, brand and key attribute in the first 70 characters, using words with real search volume.

How long can a Google Shopping product title be?

Google allows up to 150 characters in the title attribute. Shoppers usually see only the first 70 characters or fewer, so the most important details must come first. Use the rest for useful details like size, color or material, never for price or shipping text.

Do I need a GTIN for every product?

No. Google recommends a GTIN for every product that has one from the manufacturer, and a missing or wrong GTIN may limit visibility. If a product has no GTIN, submit brand and MPN where they exist, or set identifier exists to no.

Can I change product titles without editing my Shopify store?

Yes. A supplemental data source in Merchant Center can update details on existing products, including titles, without changing the primary feed from Shopify. It cannot add or remove products. I use it to test new titles on a group of products before rolling them out.

Why are my products disapproved in Merchant Center?

Common causes include a price or stock mismatch with the product page, missing identifiers, broken links, placeholder content and policy issues. Open the Needs attention tab to see each issue and its fix. Correct the data, upload again, and wait for the review.

Does feed optimization help Performance Max campaigns?

Yes. Shopping placements in Performance Max run on the product feed, so clear titles, correct identifiers and accurate categories give it better material. Custom labels also let you split products by margin, season or collection, which makes budgets and reporting easier to control.

What should you do next with your Shopping feed?

Start with the Needs attention tab, then pull your Shopping and Performance Max search terms and compare them with your top product titles. The gaps between what people search and what your titles say are usually the first fix. Then fill missing identifiers and keep price and stock synced with the site.

If you would rather hand the work to someone, you can hire a freelance Google Ads specialist who has done this across many catalogs. I hold Top Rated Plus status on Upwork, a badge for the top 3% of performers. My fee is $600 to $800 per platform a month, month to month, and you keep ownership of every account.

For more guides like this one, browse the Google Ads blog.

How this post was written: drawn from Ishant Sharma’s own account work and the published case studies linked above, researched against current Google Ads documentation, drafted with AI assistance and checked against the sources before publishing.