Start with what AI Mode shopping actually queries, because it is not your website. When someone asks for "a waterproof travel bag for a weekend trip", Gemini pulls candidates from Google's Shopping Graph, a real-time base of over 50 billion product listings refreshed continuously with price, availability, reviews and attributes.

Under the hood it runs query fan-out: the question is split into micro-intents, each hits the Shopping Graph separately, and the best-fit products come back as conversational answers, carousels and comparison tables.

So the entry ticket is your Merchant Center feed. On Shopify, that means the Google & YouTube channel, which is also how Shopify routes products to Google AI Mode and Gemini within Agentic Storefronts. No feed, no Shopping Graph. No Shopping Graph, no AI Mode shopping answer, regardless of how well your site ranks.

With the pipe connected, five things decide whether you get picked.

First, attribute completeness. Fan-out means your product is matched against clauses like "waterproof", "carry-on size" and "under $100" separately. Titles, descriptions, GTINs, variants and availability need to carry those answers, and feed specialists stress that clean, normalized data is what Gemini can actually use.

Second, reviews. Google's assistant ranks candidates using relevance, price, reviews and availability, so enroll in product ratings and keep collecting, since review volume and text shape AI recommendations.

Third, images. Features like virtual try-on require high-resolution, on-model imagery submitted through Merchant Center, and visual carousels reward multi-angle photography everywhere else.

Fourth, freshness. Price and stock sync in near real time, and products Google cannot confirm as available get dropped from answers. Fix feed disapprovals first; they are the silent killer.

Fifth, eligibility for the transaction layer. Agentic checkout is rolling out with eligible US merchants, including select Shopify merchants, built on the Universal Commerce Protocol that Google co-developed with Shopify. When the option appears in your channel settings, take it, because "Google can buy it for you" listings get the warmest intent.

Two honest caveats. Ads increasingly shape these surfaces: Google is layering sponsored formats and automated campaign signals into AI shopping, so paid and organic feed quality now compound each other. And AI Mode also answers the research questions upstream of the product, where your content, not your feed, gets cited; the difference between the two surfaces is worth understanding, and this comparison of AI Overviews and AI Mode covers it.

Give it time and watch the diagnostics. New or fixed feeds typically take days to weeks to propagate into AI surfaces, and Merchant Center's product status report tells you exactly which items are excluded and why. Appearing in the organic layer of these answers costs nothing; you are competing on data quality, not budget, which is rare enough in Google's ecosystem to exploit while it lasts.

Test like a shopper, not like an SEO. Open AI Mode, ask the questions your customers ask, with constraints, and see who comes back. If a competitor appears and you do not, diff their listing against yours attribute by attribute. The gap you find is the roadmap, and it will almost always be data, not content.