Journalist at Promodo
For over 6 years, I have been working as a journalist in the communications and marketing industry. I help brands communicate effectively through written content, engage with market experts, and create professional materials on topics related to business and marketing, sharing insights on working with marketing tools.
Since users increasingly search for products by specific attributes such as material, color, or style, businesses may miss out on relevant traffic if those details are not included in their product feed.
This was exactly the challenge the Promodo team faced while working with our clients, Etnodim and MBM. In both cases, our analysis showed that potential traffic was being lost due to limitations in the product feed structure.
We analyzed user search behavior, identified opportunities to improve product data structure, and developed a feed optimization approach that significantly improved Shopping campaign performance.
A product feed is a structured file that contains information about products and is uploaded to Google Merchant Center. Google uses this data to create product ads and determine which search queries are relevant for displaying those products.
The quality of a product feed directly impacts the performance of Google Shopping, Performance Max, dynamic remarketing, and other catalog-based campaigns. For these campaign types, the feed has become one of the main sources of signals used by Google's algorithms.
The challenge is that most product feeds are created primarily for website catalogs rather than advertising systems. That’s why product titles and descriptions often lack the attributes that users include in their searches.
For example, someone may search not just for a "women's dress" but for a "white linen dress." If these attributes are missing from the feed, Google has fewer signals to determine whether the product is relevant to the search.
This is especially important for non-branded searches, where users are not looking for a specific brand but are searching based on product features and characteristics.
So, we recommended optimizing the product feeds for our clients, Etnodim and MBM, to improve the performance of their advertising campaigns.
The optimization process included three stages:
Product feed enrichment is the process of improving and expanding product attributes to increase the effectiveness of advertising campaigns.
The Promodo team analyzed the characteristics users most frequently include when searching for products in relevant categories. We then compared these search patterns with the information available in our clients' product feeds.
This analysis helped us identify:
In essence, we developed a new approach to structuring product data. Targeted not only at website catalogs, but also at Google Shopping algorithms.
The final step was to implement these changes in the product feeds quickly, efficiently, and without requiring additional resources (such as budget or manual effort).
For small product catalogs, updating titles and descriptions manually is manageable. However, when a retailer manages thousands of SKUs, regularly updates its assortment, and runs campaigns across multiple categories, the process quickly becomes too time-consuming and resource-intensive.
Maintaining a consistent feed structure is another challenge. In large catalogs, there are often duplicate entries, inconsistent naming formats, incomplete attributes, and descriptions that don't align with the way users actually search for products.
To solve these challenges, we used AI. This approach significantly reduced the time required to implement feed updates and helped us maintain a consistent and well-structured set of product attributes.
To automate the process, we used FeedGen, an open-source solution developed by Google and Admixer Advertising. The tool uses generative AI and large language models (LLMs) to optimize product feeds.
It allowed us to apply the enrichment rules we had developed across large product catalogs at scale. Based on the defined attributes, the system generated updated product titles and descriptions and helped fill in missing product information.
In addition, FeedGen can analyze product images and identify attributes that are missing from the source data. For example, if a product's color or style is not included in the feed, AI can detect and automatically add these attributes.
Etnodim is a Ukrainian brand of contemporary embroidered clothing.
Etnodim already had strong visibility for branded searches, but there was an opportunity to attract more non-branded traffic.
Our analysis of Shopping campaigns showed that users were actively searching for products based on attributes such as material, garment type, design details, and style. However, many of these attributes were either missing from the product feed or were insufficiently represented.
We analyzed user search patterns, created title templates that combine key product attributes in a logical and easy-to-understand order, and selected two title-generation formats:
Material | Gender | Product Type | Color | Title | Brand
Product Type | Gender | Color | Title | Brand
The Promodo team used FeedGen to build a title-generation framework based on real user search patterns. The AI model generated new product titles and descriptions, added missing attributes, and structured product information to make it easier for Google to understand.
shopping feed optimization
What made this project unique was that FeedGen was used not only to improve existing product data but also to generate new information based on visual content.
As a result of the feed optimization:
While working with MBM, we faced a different challenge.
MBM is an eCommerce retailer specializing in home goods with a large product catalog. Many products on the site had incomplete attributes, and product descriptions were cluttered with technical symbols and unstructured information. As a result, Google lacked the signals it needed to accurately determine product relevance.
We analyzed the product categories and identified the attributes that had the greatest impact on search relevance. Then we developed a new structure for product titles and descriptions that provided Google's advertising algorithms with more complete and consistent product information.
The AI model generated new title variations that incorporated attributes such as material, brand, color, product type, and other relevant characteristics. At the same time, it cleaned up product descriptions and converted them into a more structured format.
As a result:
An additional benefit was the speed of implementation. More than 500 products were processed in approximately 3 hours without developer involvement or manual rewriting of product listings.
Many eCommerce teams still view the product feed as a technical component of Google Merchant Center rather than an extra tool for improving Shopping campaign performance. At the same time, the Promodo case studies show that feed optimization alone can help businesses reach more relevant search queries and improve Google Shopping results without increasing budgets or launching new campaign types.
This is especially valuable for retailers with large product catalogs, where manually updating titles, descriptions, and attributes can be time-consuming.
If your product feed needs improvement or you'd like to explore an AI-driven approach to feed enrichment, the Promodo team can help you evaluate your current setup and identify opportunities to improve performance.
Journalist at Promodo
For over 6 years, I have been working as a journalist in the communications and marketing industry. I help brands communicate effectively through written content, engage with market experts, and create professional materials on topics related to business and marketing, sharing insights on working with marketing tools.
You may also like
Choose quality and trusted services to improve the presence of your company on the Internet, and feel free to contact our UK team if you have any questions.
In this blog post, Kateryna Husachenko, Marketing Strategist at Promodo, reveals why is it important to know your customers
Polyvore is all about fashion. It helps you to sell, buy and promote.
Above 90% of your potential customers are seeking property management services like yours online. This understanding comes as a game changer.
We at Promodo are ready to help you improve your performance across all digital marketing channels.
Get started
.avif)
.svg){kind=logo}
