The Google Shopping Graph: How Product Data Powers AI Mode Shopping

Ask Google's AI Mode for "a waterproof hiking jacket under £150 that packs small" and the answer is not written from general web pages alone. It is assembled from the Shopping Graph: Google's database of products, sellers, prices, availability and reviews. Whether your products appear in that answer depends largely on how complete, accurate and current your data is in that graph.

The scale is hard to picture. Google said in May 2025 that the Shopping Graph held more than 50 billion product listings, with more than 2 billion refreshed every hour. In 2026, the same data increasingly powers checkout inside AI Mode and the Gemini app, through Google's Universal Commerce Protocol.

This guide explains what the Shopping Graph is, how Merchant Center, reviews and your own website feed AI Mode shopping answers, and which product data fields matter most, with a step-by-step plan to improve your feed.

Key Takeaways

• The Shopping Graph is Google's product database, covering large retailers and small local shops, with prices, reviews, colour options and availability.
• AI Mode combines Gemini with the Shopping Graph and uses query fan-out to work out what makes a product right for a request.
• Your main inputs are a Merchant Center feed and Product structured data on your pages, plus reviews and consistent on-site information.
• Required feed fields cover the basics; optional fields such as product_highlight, product_detail and question_and_answer add the detail conversational queries need.
• Checkout inside AI Mode and Gemini via UCP is limited to select merchants in the US, Canada and Australia and needs the native_commerce attribute.

What is the Shopping Graph?

Google describes the Shopping Graph as a dataset of products and sellers that spans everything from global brands to neighbourhood stores, with details such as reviews, prices, colour options and availability.

“The Shopping Graph now has more than 50 billion product listings”

— Google, Shop with AI Mode, use AI to buy and try clothes on yourself virtually

In the same announcement, Google said AI Mode brings Gemini together with the Shopping Graph. For browsing queries it shows a panel of products; when a shopper adds detail, it runs several searches at once to work out which attributes matter (Google's example: whether a bag is waterproof for rainy weather), and updates the product panel as the conversation narrows. For more on the AI Mode side, see our guide to Google AI Mode and our post on AI shopping visibility.

How data flows into AI Mode shopping answers

Source	What it supplies	Your control
Merchant Center product data	Structured product facts: title, price, availability, identifiers, images, attributes	Full; you submit and maintain the feed
Product structured data on your site	Product, offer, price, availability and review markup that can make pages eligible for merchant listing experiences	Full; must match visible page content
Product page content	Descriptions, specs, sizing, compatibility and FAQs	Full
Reviews and ratings	Signals about quality and fit for specific uses	Indirect; earn and display genuine reviews
Other sellers and the wider web	Competing offers and third-party descriptions	Indirect

Google's guidance on AI features says to keep Merchant Center and Business Profile information up to date, and that structured data should match what is visible on the page. Our guides to Google Merchant Center and merchant listing product schema cover the set-up.

The product data fields that matter most

All fields below are in Google's Merchant Center product data specification. The "why it matters for AI" column is our practical reading of how conversational shopping uses them.

Field	Status in the spec	Why it matters for AI shopping answers
title	Required (max 150 characters)	Should name brand, product type and key attributes the shopper will mention
description	Required (max 5,000 characters)	Room for use cases, materials and fit that answer detailed requests
price, availability	Required	"Under £150" and "in stock now" filters depend on them
image_link	Required	Product panels are visual; clear images help selection
gtin, brand, mpn	Conditionally required	Identifiers help Google match your offer to the right product
color, size, material, item_group_id	Required for some categories and variants	Conversational queries often specify these
product_highlight	Optional (2–100 highlights, 150 characters each)	Short, scannable benefits an AI answer can reuse
product_detail	Optional (up to 100 entries)	Technical specs for comparisons ("packs to 20 cm")
question_and_answer	Optional (up to 30 pairs)	Direct answers to common pre-purchase questions
native_commerce	Needed for UCP checkout	Only listings with checkout eligibility show the "Buy" button in AI Mode and Gemini

Google's spec says question_and_answer content should not repeat product_detail, product_highlight or description, so use it for genuinely new information such as care instructions or compatibility questions. For variant handling, see our product variants SEO guide.

Checkout in AI Mode and Gemini

Google's May 2025 announcement introduced agentic checkout: shoppers can track a price, and when it drops to their target, confirm and let Google buy it through Google Pay. In 2026 this has developed into the Universal Commerce Protocol (UCP), an open standard for agentic commerce. Merchant Center Help says eligible merchants can add a checkout button to listings in AI Mode in Search and on Gemini, with checkout happening on Google's surfaces while the merchant remains the seller of record. It is currently limited to participating merchants in the US, Canada and Australia, and Google's March 2026 update added multi-item carts, real-time catalogue access and identity linking for loyalty perks.

If you sell through ChatGPT as well, compare the requirements with our guide to OpenAI product feeds.

A step-by-step plan to strengthen your Shopping Graph presence

1. Fix errors first. Clear disapprovals and warnings in Merchant Center, especially price and availability mismatches between feed and site.
2. Rewrite titles for real requests. Lead with brand and product type, then the attributes shoppers ask for.
3. Fill the optional detail. Add product_highlight and product_detail to your top sellers, then extend across the catalogue.
4. Add questions and answers. Use customer service logs to write question_and_answer pairs for common pre-purchase questions.
5. Align your product pages. Make sure Product structured data, visible prices and stock match the feed. See structured data for AI search.
6. Grow genuine reviews. Ask buyers to mention use cases ("great for wet walks"); see reviews and AI recommendations.
7. Assess UCP readiness. If you sell in an eligible country, review Google's requirements and decide whether to submit an interest form.
8. Test in AI Mode. Run 20 realistic shopping prompts monthly and record whether your products appear and which attributes are mentioned.

Common mistakes

• Treating the feed as an ads-only asset. The same data powers free listings and AI Mode answers; neglecting it costs organic visibility.
• Keyword-stuffed titles. Titles should read naturally and describe the product accurately.
• Stale price or stock. With billions of listings refreshed hourly, mismatches are noticed and erode trust. See our guide to pricing information in AI answers.
• Repeating the description in Q&A. Google asks that question_and_answer add new information.
• Assuming checkout is open to all. UCP checkout is limited to select merchants in a few countries for now.

Related Guides

• Meta AI: How Brands Appear in Answers on WhatsApp, Instagram and Facebook
• Apple Intelligence and Siri: What Changes for Brand Discovery
• Alexa+ and Local Businesses: Getting Recommended by Amazon's Assistant

Frequently Asked Questions

What is the Google Shopping Graph?

It is Google's dataset of products and sellers, including prices, reviews, colour options and availability. Google said in May 2025 it held more than 50 billion product listings, with over 2 billion refreshed every hour.

How do I get my products into the Shopping Graph?

The main routes are a Merchant Center product feed and Product structured data on your product pages. Both should be accurate and consistent with what shoppers see on your site.

Which product fields matter most for AI Mode?

The required basics (title, description, price, availability, image, identifiers) come first. After that, optional fields such as product_highlight, product_detail and question_and_answer add the detail conversational queries need.

Can shoppers buy my products directly in AI Mode?

Only if you take part in Google's Universal Commerce Protocol checkout, currently limited to select merchants in the US, Canada and Australia, and your listings use the native_commerce checkout eligibility attribute.

Do I need to run Shopping ads to appear in AI Mode?

Google's documentation does not say ads are required. Organic product data from Merchant Center and your site feeds the Shopping Graph, while Shopping ads are a separate paid option.

Conclusion

In AI Mode, your product data is your shop window. Clean, complete Merchant Center feeds, matching structured data and genuine reviews give the Shopping Graph what it needs to recommend you for specific, conversational requests. Our ecommerce SEO team can audit your feed and product pages for AI shopping. Get a free quote.

References

1. Google — Shop with AI Mode, use AI to buy and try clothes on yourself virtually (20 May 2025)
2. Google Merchant Center Help — Product data specification
3. Google Merchant Center Help — Universal Commerce Protocol checkout
4. Google — UCP updates (19 Mar 2026)
5. Google Search Central — Merchant listing (Product, Offer) structured data
6. Google Search Central — AI features and your website