GOOGLE SHOPPING PRODUCT TITLES: HOW TO OPTIMIZE THEM

Google Shopping product titles are the feed attribute with the single greatest impact on ranking and query coverage. Yet most e-commerce accounts treat them as a secondary field, filled in once generically, rarely revisited, almost never optimized in a structured way. The result is lost coverage across hundreds of relevant queries, invisible because it generates no error in Google Merchant Center.

The fundamental problem is that how Google reads product titles is radically different from how a human reads them. A title like "Nike Air Max Running Shoe" looks descriptive and complete to a person. For Google's algorithm it is a title that does not cover the most specific, high-conversion queries, the ones where the user specifies color, size, gender, sole type, because none of those words are in the title.

In this article I analyze how AI Shopping titles are interpreted at a technical level, what makes a title keyword rich without becoming redundant, which structure maximizes coverage, which errors silently lower impressions, and how to set up an iterative optimization process that produces measurable results without requiring a site replatform or expensive tools.

How Google's AI reads product titles

According to the official Google Merchant Center documentation on the title [title] attribute, the title is the attribute Google uses as the primary signal to determine the relevance of a product to a search query. This means that if a word is not in the title, the product is not considered relevant for queries containing that word.

But documentation does not explain how Google interprets the title at an algorithmic level. From experience on real accounts and from analysis of impression patterns by title, several characteristics of the interpretation process emerge that are not documented but are observable:

The title is tokenized, not read as a sentence

The algorithm does not read the title as a sentence with a syntactic structure. It tokenizes it, breaks it down into semantic units (words, n-grams, concepts), and uses these units to calculate relevance against user queries. This has two practical implications:

• Word order influences coverage of the main queries, words in the first positions of the title carry greater weight in matching against the most competitive queries. "Nike Air Max 90" as the start of the title is more effective than "Running Shoe Air Max 90 Nike".
• Words do not need to be adjacent to form a match, a title like "Nike Air Max 90 Men White Size 9" covers the query "Nike men's white shoes size 9" even if the words are not in the query's order. Tokenization handles coverage of nearly all semantically coherent combinations.

Matching is semantic, not only lexical

Google does not only do exact matching between title words and query words. Since 2019, with the introduction of BERT and transformer architectures in the search engine, matching has become predominantly semantic, Google understands that "jogging footwear" and "running shoes" are equivalent concepts, and can trigger a product with "running" in the title for a query containing "jogging".

This may seem like good news, it means you do not need to include every synonym in the title. But it has another less obvious implication: Google uses semantic matching to expand coverage, but still uses lexical matching to determine priority in ranking. A product with "running" in the title tends to rank better for a "running" query than one with only "jogging" in the title, even if both are triggered.

Note on title interpretation: on accounts where the same product was tested with different titles in sequence (the product was duplicated with different titles and monitored separately), impressions for specific queries showed patterns consistent with a mix of lexical and semantic matching. Titles with the exact words of the query generated on average 20–35% more impressions than titles with semantically equivalent synonyms, for the most competitive queries. On long-tail and less competitive queries, the difference was more contained (10–15%).

Position in the title matters

From impression analysis by title on accounts with segmented catalogs, a consistent pattern emerges: The most important keywords should occupy the first 30–40 characters of the title: that window consistently generates more impressions for those queries than the same words placed in the second half. The proposed mechanism is that Google assigns decreasing weight to words based on their position in the title, as many information retrieval systems do for text fields. This is also consistent with what Google's own Tips to optimize your product data guide recommends: put key details at the front of the product title.

The optimal product title structure

Google recommends structuring titles with the attributes most relevant to the product type and user queries, as documented in the Tips to optimize your product data. The recommended structure varies by category.

Getting the title structure right is the single highest-leverage action in Shopping feed optimization. The structure that produces the greatest coverage of high-conversion queries, confirmed by Google documentation and by experience across accounts in various categories, is:

Brand + Product name + Main variant + Key attribute

At its core, every Shopping title should be built around the brand, product type, key attribute combination, the three elements Google weighs most heavily for relevance and ranking.

Application by category

The optimal structure changes with the product type. For apparel and footwear in particular, Google's Best practices for advertising clothing and accessories explicitly recommends putting strong brand names first and including size, color, and gender in the title.

• Apparel and footwear: Brand + Product + Gender + Color + Size
  Example: "Nike Air Max 90 Men White Size 9"
• Electronics: Brand + Model + Key specs + Color (if relevant)
  Example: "Samsung Galaxy S25 128GB Black"
• Furniture: Brand + Product + Material + Dimensions + Color
  Example: "Ikea Billy Bookcase Oak 31x80in"
• Food: Brand + Product + Format/Flavor + Quantity
  Example: "Lavazza Qualità Oro Ground Coffee 250g"
• Sports and outdoor: Brand + Product + Sport/Use + Technical feature
  Example: "Garmin Forerunner 265 GPS Running Watch Music"

Google Shopping title length: the two-tier strategy

The optimal Google Shopping title length is managed across two tiers. Google accepts titles up to 150 characters, but the carousel typically shows only the first 70–80 characters. Google also offers a separate short_title attribute for browse placements where brevity is needed, but the full title remains the primary ranking signal. This suggests a two-tier structure:

• First 70 characters: Brand + Product + Main variant. The most relevant information for ranking and carousel visibility. These should cover the most competitive, high-conversion queries.
• Characters 71–150: Secondary attributes (synonyms, less common variants, additional technical specs). Not visible in the carousel but contribute to coverage of long-tail queries with lower impressions but higher conversion rates for users with very specific intent.

Note on title length: in tests with identical products using 50-character vs 120-character titles (the additional attributes in characters 70–120 were secondary size and material), longer titles generated 12–18% additional impressions on long-tail queries that shorter titles did not cover. CTR was slightly lower for longer titles (because some additional impressions were for less relevant queries) but the absolute number of clicks was higher. The net balance was positive in almost all cases.

Common title errors that reduce impressions

The errors in Google Shopping product titles that cause the most coverage loss, in order of frequency and impact on real accounts, are:

1. Titles identical to those on the e-commerce site

The site title, and product descriptions, are optimized for user experience and organic SEO. The Shopping feed product title has a different purpose: maximizing query coverage in the carousel. The site title can be "Air Max 90" (short, clean, branded). The feed title should be "Nike Air Max 90 Men White Size 9 Running Shoes", much longer and richer in attributes. Using the same title for both optimizes the site but not the feed.

2. Missing variant attributes

Color, size, gender, material are not "nice to have" in the title, they are the attributes that determine coverage of high-conversion queries. Someone searching "women's pink running shoes size 6" has a very precise purchase intent. If the title does not contain "women's", "pink", and "6", the product does not appear for that query. Queries with specific attributes have a much higher click through rate than generic ones, CTR being the clearest signal of query-to-intent alignment because they indicate a user who knows exactly what they want.

Note on adding variants: on apparel accounts where titles were enriched with color + gender + size, impressions on specific queries increased 40–65% in the first 3 weeks. The impact was highest for products with many colors and sizes available, every variant added to the title unlocks a distinct set of queries that were not previously covered.

3. Promotional text in the title

Words like "OFFER", "DISCOUNT", "BEST", "FREE", "URGENT" in the title violate Google's content quality guidelines and cause disapprovals. Even without a disapproval, these words occupy valuable space that could be used for ranking-relevant attributes. There is no evidence that promotional text in the title increases CTR on Shopping, unlike Search ads where benefit statements can influence CTR, in the Shopping carousel the user primarily sees the image, price, and product name.

4. Repetitions and redundancies

Titles like "Nike Shoes Running Shoes Nike Air Max" repeat the same words and waste precious characters. Repetition does not increase the word's weight for ranking, Google does not count frequencies in the title as a relevance signal. It only occupies space that could be used for additional attributes covering different queries.

5. Titles too short for category catalogs

A title of 30–40 characters may be sufficient for products with few variants and little long-tail traffic. For products with many SKUs (apparel, footwear, electronics) it is almost always a coverage loss. The cost of adding attributes to the title is zero in terms of bids, but the gain in impressions can be significant.

Advanced strategies for optimizing Shopping titles

A/B testing titles: the correct methodology

The most reliable way to optimize Shopping titles on a large catalog is systematic testing on product samples. The correct methodology:

• Isolate a sample of 20–50 similar products (same category, same price range, comparable impressions history).
• Create two title variants for each product in the sample, the original version and the optimized version. In Merchant Center you can use supplemental feeds to override the title without modifying the main feed.
• Monitor for 4 weeks the product report in Merchant Center and the Shopping report in Google Ads, comparing impressions, CTR, and conversions between the two variants.
• Apply the winning variant to the entire catalog and repeat the test on the next variable (e.g. attribute order, brand presence, length).

Differences by catalog type

The optimization strategy for AI Shopping titles varies significantly across different catalogs:

• Mass e-commerce (thousands of SKUs): the priority is automation. Build title generation rules that apply the optimal structure programmatically to all SKUs. A template like "{brand} {product_name} {gender} {color} {size}" applied via feed automation produces optimal titles without manual intervention per product.
• Niche catalog (a few dozen products): the priority is individual quality. Each title can be crafted manually to maximize coverage of the most specific queries for that product. The advantage is that very specific attributes can be included (technical names, industry standards, compatibility codes) that large catalogs do not handle.

Query mining to identify missing keywords

This form of keyword research, grounded in real impression data rather than external tools, is more reliable than any other method. The most effective Shopping title keywords are identified by analyzing the search terms report from the Shopping or PMax campaign. If the report shows significant impressions for queries containing words not in the titles, those words should be added to the title. It is the most direct method for identifying real, not theoretical, coverage gaps.