How Does Google Business Agent Use Merchant Center Data, and Can Weak Product Feeds Make a Business Invisible?

How does Merchant Center data flow into Google's AI?

Google's own guide to its generative AI features says that using Merchant Center feeds and Business Profiles "can help your products and services to be visible in both AI responses and other Google Search results". It also points merchants to Business Agent.

What is Business Agent and what does it need?

According to Merchant Center Help, Business Agent is "a conversational experience on Google Search that allows shoppers to chat with your brand", acting like a virtual associate. It grew out of a chat feature added to merchant brand profiles in 2025 in the US, Canada, Australia and India, which answered questions from Merchant Center data. Business Agent is rolling out in the US and adds your website as a source.

To customise it, a business needs a US-based online store, a verified Merchant Center account and a claimed brand profile on Google Search. Merchants set the tone of voice, welcome message, colours and a support contact, either in Merchant Center under Marketing or through the Google & YouTube app on Shopify. Until you customise it, shoppers see a standard Google-branded version. Google says training on further business data, such as size guides, and agentic checkout are coming.

The agent can only be as good as its sources. If the feed lacks sizes, materials or compatibility, and the website does not state them, it cannot answer those questions well.

Do LLMs use Google Ads and Merchant Center feeds?

Google's AI features do use Merchant Center data, through the Shopping Graph. Google describes the Shopping Graph as more than 50 billion product listings, 2 billion of which are updated every hour, and says AI Mode and shopping in the Gemini app are powered by it. Shopping ads, run through Google Ads using Merchant Center products, can also appear in AI Mode and AI Overviews, labelled "Sponsored".

Other assistants do not read your Google Merchant Center account directly, but some can use the same data. Microsoft's Google Import copies your Merchant Center products into Microsoft Merchant Center, whose feeds help inform organic Copilot shopping results, and OpenAI accepts Google-compatible feeds. Others may only see its effects. Semrush found ChatGPT's top product was in Google Shopping's top three in 75% of runs, and Search Engine Land reported that 83% of ChatGPT carousel products came from Google Shopping data. How ChatGPT ranks products is covered in how ChatGPT ranks shopping results. There is no evidence that any assistant reads your Google Ads account settings or bids.

How do Google Shopping Graph and Merchant Center affect AI product recommendations?

The Shopping Graph combines merchant feeds with information Google finds across the web, such as product pages, reviews and inventory. When someone asks AI Mode a detailed shopping question, the answer is built from that data, with prices, reviews and stock. That makes feed quality a visibility issue:

• Disapproved products cannot appear in free listings or ads.
• Missing identifiers, such as GTINs, make it harder to match your listing to a known product and its reviews.
• Thin attributes leave nothing to match detailed questions like "waterproof hiking boots with a wide toe box".
• Stale price or stock creates mismatches, and can lead to disapproval.

Merchant Center also announced AI performance insights in May 2026, to show how products are discovered on AI Mode, AI Overviews and the Gemini app, and made it available in September 2026 in the US, Canada, Australia, India and New Zealand.

Do Google Merchant Center free listings appear in Gemini product answers?

Google says the Gemini app now offers shopping features "powered by Shopping Graph information", with shoppable product listings, comparison tables, prices and places to buy. Merchant Center's AI performance insights cover the Gemini app alongside AI Mode and AI Overviews. So products in Merchant Center, including free listings, can be shown in Gemini's shopping answers. Google has not published how it chooses which listings to show there. Checkout through Google's Universal Commerce Protocol is available to selected merchants on eligible listings in AI Mode and the Gemini app. Checkout inside AI Mode and Gemini has its own feed and account requirements, covered in what Google's UCP setup requires of Merchant Center sellers. Paid deals shown in the same answers are covered in whether Direct Offers affect organic recommendations.

How do I write a structured_description for Google Merchant Center without hurting product visibility in AI search?

Merchant Center requires descriptions written with generative AI to be submitted in the structured_description attribute. It has two sub-attributes: digital_source_type, set to trained_algorithmic_media for AI-generated text, and content, the description itself. AI-written titles go in structured_title in the same way. If a person wrote the description, the ordinary description attribute is fine.

Flagging the text is a disclosure requirement, not a ranking penalty that Google has described. What affects visibility is what the description says. Good practice:

• Start with facts: what it is, key specifications, materials, sizes and compatibility.
• Answer real questions shoppers ask, in plain language.
• Check every claim against the product. AI tools invent features, and errors can become disapprovals or wrong answers.
• Avoid promotional filler, such as "best ever" or "free shipping", which Google's spec discourages.
• Keep it consistent with the product page, so Google sees the same facts in both places.

Google Merchant Center feed versus Product schema: which source do AI assistants trust when product data conflicts?

For Google, the landing page can override the feed on some facts. Merchant Center's automatic updates, on by default, read structured data and other data on your product pages and update price, sale price, availability and condition in free listings and ads. Google's example: if the feed says $4 and the page says $3, it updates the listing to $3. But Google says the product may be disapproved instead if it detects a mismatch, and it may stop updating if mismatches become too many. Automatic updates are meant for occasional errors, not as a replacement for regular feed updates.

For other assistants, there is no published rule. ChatGPT reads both merchant feeds, where submitted, and crawled pages. The safe answer is not to let them conflict. Generate the feed and the Product schema from the same source, and update both when prices or stock change. Conflicting facts are one reason AI answers get prices wrong, as covered in why Google AI Overviews hallucinate prices or business details.

Does Google AI Mode cite Merchant Center product data or only web pages?

Both, in different ways. AI Mode's shopping responses show product listings with prices, reviews and stock drawn from the Shopping Graph, which includes Merchant Center data. These appear as product cards, not as cited web pages. The text of the answer is supported by links to web pages, found through Google's search index and query fan-out. So a retailer can appear as a product listing without its site being cited, or be cited as a page without its product appearing. Strong visibility needs both: a complete feed and product or guide pages that answer the question.

What should you fix first?

1. Clear disapprovals and warnings in Merchant Center diagnostics.
2. Add identifiers: correct GTINs, brand and MPN.
3. Complete attributes buyers filter by, such as size, colour, material and compatibility.
4. Add conversational attributes where useful. Google lists question and answer, document link, related product, item group title, variant option and popularity rank.
5. Keep price and stock in sync between the feed and the page.
6. Claim your brand profile and set up Business Agent if you are eligible.

How much of this depends on your wider presence is covered in whether a weak external footprint can make a brand invisible to AI assistants.

We have not run a Merchant Center account ourselves for this page; it is based on Merchant Center Help, Google's announcements and published studies.

What might this look like in practice?

Consider an illustrative outdoor gear shop, Ridgeway Supply, whose hiking boots rarely appeared in AI Mode answers for questions about wide feet. Its feed used short titles, had no GTINs for 40% of products, and described width only on the product pages. The team added GTINs, put width and waterproofing into the feed's product details, and added two question-and-answer pairs per boot using the new attribute. They also rewrote AI-generated descriptions, checked them against the boots, and moved them to structured_description. They claimed the brand profile and customised Business Agent. Within weeks, several boots began appearing in AI Mode for wide-fit questions.

What do people ask about Business Agent and Merchant Center?

Who can use Google's Business Agent?

US-based online stores with a verified Merchant Center account and a claimed brand profile.

Does Gemini use Merchant Center data?

Yes. Google says shopping in the Gemini app is powered by the Shopping Graph, which includes Merchant Center listings.

Do I have to flag AI-written product descriptions?

Yes. Merchant Center requires them in the structured_description attribute with digital_source_type set to trained_algorithmic_media.

What happens if my feed price differs from my page?

Google may update the listing from the page, or disapprove the product if it detects a mismatch.