Someone types “running shoes for flat feet under $120” into Gemini. Three products come back, with prices, stock status, and a line on why each one fits the ask. Gemini didn’t crawl a single page to build that answer. It queried the Shopping Graph, the same structure it taps for Search and YouTube results. Your product’s visibility in that answer has nothing to do with how well-formatted your feed is. It has everything to do with whether your product has a strong entry in that graph already.
Key Takeaways
- Gemini’s shopping answers draw from the Shopping Graph, a continuously updated structure of listings, pricing, reviews, and availability, not from live web crawling at answer time.
- A clean product feed gets you indexed into the graph. It does not determine whether Gemini selects and names your product in an answer.
- Shopping Graph presence and general AEO work overlap but are not the same task: one is retrieval into a structured commerce layer, the other is retrieval into open-web and document-based answers.
What Gemini Actually Queries
Gemini already describes itself as pulling from Search, YouTube, and Scholar. For shopping, add a fourth: the Shopping Graph. Google’s standing record of what exists, what it costs right now, who sells it, and what buyers said about it. This isn’t new infrastructure. It’s the same backbone running Google Shopping and the shopping panels in regular Search results. Gemini just plugged a conversational layer on top.
Here’s the part people get wrong. The graph isn’t assembled fresh for your query. It’s a persistent index, built from merchant feeds, retailer sites, review platforms, and structured markup, then reconciled down to one entry per product. When Gemini answers, it’s picking from that standing index. Not searching the live web. No entry means nothing to pick.
Your Feed Is One Input. The Graph Is the Whole File.
A product feed submitted through Merchant Center tells Google what you sell and what it costs. That’s it. The feed doesn’t aggregate reviews. It doesn’t track whether three other retailers carry the same item at a different price. It doesn’t confirm your listing details against independent sources. The graph does all of that, and it’s the graph Gemini actually consults.
Two brands can submit equally clean feeds and land in completely different places. Brand A has reviews scattered across its own site. Brand B has reviews, specs, and pricing confirmed on its own site, two marketplaces, and a handful of independent review outlets. Brand B’s entry is denser. Better corroborated. Gemini leans toward surfacing it, because the graph runs on agreement across sources. Same mechanism that governs trust everywhere else in AEO.
growth in AI brand presence for NewDay USA, tracked weekly across five engines over six months.See the case study
Ranking Well and Getting Cited Are Different Games
A brand ranks page one for “best running shoes for flat feet.” Gets reviewed by a dozen publications. Runs a feed with thousands of SKUs in good standing. That’s Share of Voice, and none of it guarantees Gemini names the brand when a shopper asks directly. That’s Share of Citation. In shopping, citation gets earned by the graph entry. Not by general web presence.
Share of Voice measures how often you are mentioned. Share of Citation measures whether Gemini pulls your product into the answer and names you as the source for it.
The gap shows up constantly in shopping categories, mostly because so much of what Gemini weighs (price accuracy, stock status, cross-retailer agreement) sits outside normal content marketing entirely. A brand can dominate search results and still disappear from the graph’s answer because its pricing is stale in three of five places it’s listed. Ranking well doesn’t fix stale pricing. Only the graph entry does.
Where Feed Optimization Stops Short
Feed optimization answers one question: is this product described correctly and indexable. It says nothing about whether the product is corroborated across the sources Google reconciles into the graph. You can structure a feed perfectly and still end up with a thin entry if the product has zero third-party reviews, zero marketplace listings, and no price history beyond your own site.
Structured data on the product page helps the feed. It does almost nothing for the graph’s trust signals. What matters there is entity consistency: same product name, same specs, same identifiers, everywhere the product shows up. That consistency is what lets Google merge listings into one strong entry instead of three weak, conflicting ones. Sell the same shoe under slightly different names on your site, a marketplace, and a retail partner, and you’ve handed Google three weak signals where one strong one should be.
Building an Actual Graph Presence
This is its own workstream. Separate from feed hygiene, with its own checklist.
Keep product identifiers (GTINs, MPNs, brand names) identical across every channel, marketplaces and resellers included. Go after reviews on third-party platforms, not just your own site. Cross-source corroboration is what strengthens the entry. Keep pricing and availability current everywhere the product is listed; gaps between sources weaken confidence in the whole entry. Treat retailer and marketplace listings as citation sources worth auditing, the same way you’d audit backlinks.
None of this replaces feed optimization. It runs alongside it, the way structured data and digital PR run alongside each other in broader AEO work. One makes a product retrievable. The other makes it trustworthy enough to get chosen.
What to Check This Week
Start by asking Gemini the direct shopping questions a real buyer in your category would type. Do your top sellers show up? If not, pull listings from your three or four biggest retail partners and compare prices, names, and specs against your own site. Inconsistencies there are usually the first thing dragging your graph entries down.
Then look at where reviews for those products actually live. All on your own domain? That’s a corroboration gap, and it’s worth closing before you spend another dollar on feed cleanup alone.
FAQs
Is the Shopping Graph the same thing as a Merchant Center feed?
No. The feed is one input Google uses to build graph entries. The graph itself also incorporates reviews, cross-retailer listings, and price history, reconciled into a single record per product.
Does better feed optimization guarantee Gemini will cite a product?
No. A clean feed makes a product indexable into the graph. Whether Gemini selects and names it in an answer depends on how corroborated and current that graph entry is across sources, not on feed formatting alone.
What is the difference between Share of Voice and Share of Citation for shopping queries?
Share of Voice is how often a product or brand is mentioned across the web generally. Share of Citation is how often Gemini actually pulls that product into a shopping answer and names it as the source. A brand can be strong on one and near-zero on the other.
Do third-party reviews matter more than on-site reviews for Shopping Graph visibility?
Third-party and marketplace reviews add corroboration that on-site reviews alone cannot provide. A graph entry backed by agreement across independent sources is generally stronger than one built entirely from first-party content.
