Why your product data now reaches far beyond Shopping Ads

When the product feed served a single surface, errors stayed invisible

The historical model was straightforward. A feed uploaded to Merchant Center powered Shopping Ads. The required attributes were well-known: title, price, availability, link, image. As long as those fields were populated, ads ran.

Optional attributes were routinely ignored. Product descriptions, custom labels, material attributes, sizing, compatibility. These fields felt secondary because they had no visible impact on Shopping campaign performance.

What changed is the number of surfaces consuming that data. And each new surface values different attributes.

What Google actually consumes from your feed in 2026

The Merchant Center feed no longer serves Shopping Ads alone. Here are the surfaces that pull from it directly.

AI Mode in Search. When a user asks a conversational question like "what waterproof hiking boots work for Iceland under 150 euros," Google's AI parses structured product data to build its response. It does not search for keywords in titles. It checks attributes: material, waterproofing, price range, availability. If those attributes are missing from your feed, the product is not surfaced.

Demand Gen with product feed. Demand Gen campaigns can integrate a Merchant Center feed to turn video and image ads into shoppable storefronts across YouTube, Discover, and Gmail. Google reports that on average, advertisers who add a product feed to Demand Gen see a 33% lift in conversions. But this feature requires square images (1:1 ratio), complete descriptions, and strict consistency between the feed and the landing page.

YouTube Shopping. Ads on YouTube Shorts and connected TV are becoming shoppable through product carousels fed by the product feed. The vertical format demands adapted visuals that most legacy feeds do not contain.

Google Images. Shopping Ads now appear individually in the Images tab on mobile. Product image quality and relevance become a direct visibility factor.

Google Lens. Visual search lets users photograph a product and receive matching Shopping results. Matching relies on structured feed data, not traditional SEO.

UCP (Universal Commerce Protocol). Since January 2026, UCP allows Google's AI agents to initiate checkout directly from AI Mode or the Gemini app. The Merchant Center feed serves as the vocabulary for these agents. If the data is incomplete or inconsistent with the site, the transaction fails.

Why feeds that worked yesterday break today

A feed designed for Shopping Ads followed keyword logic. The title contained targeted search terms, the description was often minimal, optional attributes were skipped.

On conversational surfaces (AI Mode, Business Agent, UCP), the logic is reversed. AI does not look for keywords. It interprets structured attributes to answer open-ended questions. A Shopping-optimised title like "Men's Running Shoe Nike Air Max 90 Black Size 10" performs well against a Shopping query. But it is useless when an AI agent is trying to answer "what are the best running shoes for overpronators with good cushioning."

The core difference: Shopping Ads match queries to titles. AI surfaces match intent to attributes.

The most common failure cases observed in production:

The feed contains a rich title but no description. AI Mode lacks enough context to propose the product in a conversational response.

Images are landscape-only. Demand Gen and YouTube Shorts require square or portrait. The product is not eligible.

Feed price does not match the price displayed on the product page. UCP rejects the transaction because the agent detects an inconsistency.

Availability attributes are not synchronised in real time. A product shown as "in stock" in the feed but out of stock on the site destroys agent trust.

Conversational fields introduced by Google in January 2026 (product FAQ, compatible accessories, substitutes) are not populated. The product remains invisible on Business Agent and AI Mode.

What AI agents cannot infer from your data

An AI agent does not fabricate information. If it cannot find a structured attribute in the feed, it will not deduce it from the title or free-text description. Here is what an agent cannot infer:

Cross-product compatibility. If you sell phone cases, the agent will not know which models they fit unless the attribute is explicitly populated.

Seasonal relevance. A "spring-summer collection" label is not interpreted as such by an agent that sees no seasonality attribute.

Product-specific return conditions. The agent reads the global return policy from the feed but not per-product exceptions unless they are structured.

Margin or profitability. Agents optimise for conversion probability, not merchant profitability. Without value signals (custom labels, profit data), the agent favours high-converting products regardless of margin.

The trade-off every e-commerce business must make now

The question is no longer "should I enrich my feed." The question is "what level of investment in product data justifies the number of surfaces where it will be consumed."

Scenario 1: minimal enrichment. You maintain your legacy Shopping feed. Shopping Ads keep running. But your products are progressively absent from AI Mode, Demand Gen, YouTube Shopping, and UCP. Visibility concentrates on a single surface that is losing attention share to conversational formats.

Scenario 2: structural enrichment. You invest in attribute completeness, multi-format image quality, real-time synchronisation, and Merchant Center's new conversational fields. Your products become eligible across all surfaces. Enrichment cost is high, but the feed becomes a multi-channel asset.

Scenario 3: automated enrichment. You use a feed enrichment tool (like automated product feed enrichment for multi-surface visibility) to automatically generate missing attributes, adapt image formats, and maintain site-feed consistency. Marginal cost per SKU decreases, multi-surface eligibility increases.

The choice depends on catalogue size, price and stock update frequency, and market coverage. But waiting is not a viable strategy. Conversational surfaces are already capturing traffic that no longer flows through Shopping Ads.

What breaks first at scale

Product data problems are not the same at 500 SKUs and at 50,000.

On small catalogues, manual enrichment is feasible. Every listing can be reviewed. Inconsistencies are caught quickly.

On large catalogues, problems multiply invisibly. Price changes that do not propagate fast enough. Out-of-stock items that remain "in stock" in the feed for hours. Mass-generated descriptions that are identical across hundreds of variants. Images that do not meet the ratio requirements of certain surfaces.

The product feed has become infrastructure. And like all infrastructure, it degrades under load if it is not automated.

How to assess whether your feed is ready for AI surfaces

Before changing anything, diagnose. Here are the questions to ask:

How many of your products have descriptions over 100 words? On conversational surfaces, short descriptions reduce the probability of being selected by the AI.

Are your images available in square format (1:1)? This is the required format for Demand Gen and YouTube Shopping.

Are your feed's price and availability synchronised with your site within a 4-hour window? This is the threshold beyond which UCP agents refuse to validate a transaction.

Are your conversational attributes (product FAQ, accessories, substitutes) populated in Merchant Center? Since January 2026, these fields are consumed by Business Agent and AI Mode.

Do your custom labels include margin or business priority signals? Without these, Google's algorithms optimise for conversion volume, not for your profitability.

From a generative search perspective, understanding the expanded role of Google Merchant Center across the advertising ecosystem is a prerequisite for any e-commerce team operating in 2026.

Key takeaways

The product feed is no longer a technical file uploaded to Merchant Center. It is an infrastructure layer powering all of Google's commercial surfaces.

Attributes that were optional for Shopping Ads have become prerequisites for AI Mode, Demand Gen, YouTube Shopping, and UCP.

Conversational surfaces match intent to structured attributes, not queries to titles.

An incomplete feed no longer just means lost impressions. It makes products invisible to AI agents.

Automated feed enrichment is the only scalable approach to maintaining multi-surface eligibility on large catalogues.

Product data is the new limiting factor of e-commerce performance. Not the media budget. Not the bidding strategy.

FAQ

Does the Merchant Center feed really power all Google surfaces in 2026?

Yes. Shopping Ads, Performance Max, Demand Gen (YouTube, Discover, Gmail), Google Images, Google Lens, AI Mode in Search, Business Agent, and UCP checkout all consume data from the Merchant Center feed. Each surface uses different attributes with specific format and completeness requirements.

What does the Universal Commerce Protocol (UCP) change for e-commerce businesses?

UCP is an open standard launched in January 2026 that allows Google's AI agents to execute transactions directly from AI Mode and Gemini. The Merchant Center feed serves as the foundation for these transactions. If feed data is inconsistent with the site (price, stock, return policy), the agent refuses to complete the purchase.

Which feed attributes matter most for AI Mode?

Structured attributes like material, compatibility, sizing, price range, and the conversational fields Google introduced (product FAQ, compatible accessories, substitutes) are decisive. Shopping-optimised titles alone are no longer sufficient on conversational surfaces.

Does Demand Gen require a different feed than Shopping?

No, it uses the same Merchant Center feed. But Demand Gen requires square images (1:1 ratio), more complete descriptions, and explicit activation in campaign settings. A feed optimised solely for Shopping may not meet Demand Gen requirements.

How do I know if my feed is ready for Google's AI surfaces?

Check description completeness (over 100 words), square image availability, price/stock synchronisation with your site (under 4 hours lag), and whether the new conversational attributes in Merchant Center are populated. For AI-driven discovery, understanding improving Google Shopping visibility through feed optimisation helps clarify what has changed. Similarly, Performance Max campaign optimisation and the role of the product feed connects data quality directly to campaign results. For cross-platform perspective, feed-driven advertising beyond Google with Meta Shopping campaigns shows that feed quality is a universal requirement.

‍