The Shopping Graph, GTINs, and the Product Card: why your product data is a competitive weapon
Understand the corner stone of Google's Shopping Graph and why it plays such an important part for product discovery.
I've been having the same conversation with clients fairly regularly over the past year, and it usually starts with a version of the same question: why is that brand showing up more prominently in Google Shopping when their site has less traffic and their reviews are worse?
The answer, almost every time, is the Shopping Graph. This determines whether their product data is structured in a way the Shopping Graph can work with. Remember, search engines love certainty and trust.
This piece is longer than my usual articles because the topic requires some foundation before the practical implications make sense. If you've read the earlier pieces in this series on entity trust and structured data, some of this builds on concepts you already know. If you haven't, it works as a standalone entry point.
What the Shopping Graph actually is
Google's Shopping Graph is the product knowledge database that sits underneath Google Shopping, product knowledge panels, and increasingly the AI-powered search features that are changing how consumers find and verify products.
The important thing to understand about the Shopping Graph is what it isn't. It isn't a catalogue of keywords. It isn't a collection of product listings ranked by bid. It's a knowledge graph of product entities. It’s Google's attempt to understand what products actually are, not just what they're called.
When Google indexes a product it's trying to build a canonical entity: a single authoritative representation of that product that aggregates information from multiple sources. Product specifications. Price history across sellers. Reviews from the brand's own site, from third-party review platforms, from Google Shopping itself. Availability. Delivery times. The manufacturer. The brand. All of it connected to a single product entity. Not to a URL. Not to a keyword. To the product itself as a thing in the world.
This distinction matters because it changes how Google thinks about competing listings. If you and an unauthorised seller are both listing the same product, Google isn't treating those as two separate keyword matches. It's treating them as two sellers of the same product entity. It knows attributes about that entity which it then uses to decide which to surface, how prominently, and with what information attached.
If an AI shopping agent had to recommend a retailer for a product entity (GTIN) then it will recommend the one who has the most verifiable trust and information such as material, weight, shipping, returns, reviews, etc. It’s common sense.
Who controls the canonical entity for your products is, I'd argue, one of the most underexplored competitive questions in fashion ecommerce right now.
GTINs are the spine of the graph
The Shopping Graph connects product listings to canonical entities using identifiers. The primary identifier is the GTIN or the ‘Global Trade Item Number’ otherwise known as the barcode. EAN-13 in most of Europe, UPC-A in North America, variants for smaller items. For most fashion brands the relevant format is EAN-13.
When two sellers list the same GTIN, Google knows they're selling the same product. It can aggregate their data including prices, reviews, availability, delivery times, returns, etc against the same canonical entity. When a brand's Merchant Center feed includes valid GTINs and their on-site Product schema includes the same GTINs, Google can connect the brand's website directly to the entity in the Shopping Graph. Reviews on the brand's site aggregate to the entity. Price history from the Merchant Center feed populates the entity. The brand's product imagery and specifications become the primary data source for the entity.
Without a valid, GS1-registered GTIN, none of this connection happens. The product listing exists in isolation. It can't be matched to a canonical entity. Reviews don't aggregate to anything. Price history doesn't build. The product exists in Google's system as a fragment rather than a connected entity and fragments don't win competitive positioning in a knowledge graph.
For fashion brands specifically, I see two gaps consistently.
The first is coverage. GTINs need to be assigned at the variant level not just per product. This means every variant requires a unique GTIN. A jacket that comes in four colours and five sizes has twenty variants, each needing its own GTIN. Most brands have reasonable coverage on older core styles but thin coverage on newer products and almost none on their most recent drops. I've written about this in the context of pre-drop preparation. The GTIN gap on new SKUs means new products enter the Shopping Graph as unidentified entities from day one, which is exactly the wrong moment to be invisible.
The second gap is verification. Having a GTIN field populated in your schema isn't the same as having a GS1-registered GTIN that resolves correctly to your brand. Google cross-references GTINs against the GS1 registry. A GTIN that resolves to a generic entry, a different company, or nothing at all isn't just useless, it's an inconsistency that reduces the confidence Google assigns to your product entity. Run your GTINs through GS1's Verified by GS1 tool and check they resolve to your brand before you trust them in your schema.
The Product Card as competitive battleground
The practical output of all of this is the Product Card. The product card is the panel Google builds when a consumer searches for a specific product. They appear at the top of search results for product queries, showing a product image, a price range across sellers, aggregate star ratings, and a where-to-buy section listing multiple sellers.
The Product Card is built from the canonical entity in the Shopping Graph. The brand with the most complete, consistently structured product data has the most influence over what that card shows. Their images tend to be used as the primary visual. Their product specifications form the basis of the card's descriptive content. Their reviews aggregate correctly to the star rating displayed.
For a fashion brand this matters in two competitive contexts.
The first is against authorised multi-brand retailers. If your product is stocked at three retailers as well as your own DTC site, all four appear in the where-to-buy section of your Product Card. The brand with the strongest GTIN data and the most complete schema has the most control over how the card is presented. It also governs which images are shown, which specifications are displayed, and what the aggregate rating reflects. A brand that has done this work properly looks authoritative in its own product card. One that hasn't looks like one option among several with no obvious reason to prefer it. Remember, if your brand owns the GTIn then you should automatically hold that advantage.
The second context is against counterfeit and unauthorised sellers. A counterfeit listing with a fabricated GTIN doesn't resolve to your brand in GS1. Google detects the inconsistency. That seller's listing either doesn't connect to the canonical entity for your product making it an isolated fragment rather than a connected result or it connects but registers as a low-confidence source. Either way your Product Card isn't being built from their data. It's being built from yours. This is the mechanism I've described elsewhere in this series as the data fidelity gap. The GTIN is where that gap becomes most concrete and most verifiable.
There's a review aggregation point worth making specifically. Reviews in the Shopping Graph aggregate to the canonical product entity via GTIN. Reviews on your own site, marked up with Review schema on product pages with valid GTINs in their Product schema, connect to the same entity as your Google Shopping reviews. Over time, across multiple drops, this builds an aggregate rating attached to your product entity in the graph and not just to your website, but to the product itself. A brand that has been doing this consistently across twelve to eighteen months has a review depth in the Shopping Graph that a counterfeit operation or an unauthorised seller can't replicate.
Where AI shopping agents fit in this
I want to be clear about what I know versus what I'm inferring, because this is an area where the specifics aren't fully public.
What's known is that AI shopping agents like Perplexity, ChatGPT, and Gemini with shopping features, Google's AI Overviews use product knowledge graphs as part of their data sources when making product recommendations. The Shopping Graph is Google's primary product knowledge infrastructure.
What I'd argue is that the entity trust signals that matter for the Shopping Graph are the same signals that matter for AI agent product resolution. GTIN verification, canonical entity quality, review depth, brand registration data. An AI agent trying to determine which seller is the authoritative source for a product is making the same judgement Google's Shopping Graph makes when building a Product Card. The data foundation is the same. The trust threshold may actually be higher for AI agents, which I've written about earlier in this series, because they default to ambiguity rather than make an uncertain recommendation and an ambiguous product entity is one an AI agent won't confidently recommend at all.
The practical implication is that investing in Shopping Graph entity quality isn't just a Google Shopping play. It's the data foundation that increasingly matters across every discovery channel that uses product knowledge to make recommendations.
The sequence that matters
For a fashion brand taking this seriously, the work happens in a specific order.
Register GTINs with GS1 for your full catalogue, at variant level. Check that they resolve correctly against Verified by GS1. Map them into your Product schema. Keep coverage current as new products launch and GTINs assigned and verified before launch, not as an afterthought after.
Make sure your Merchant Center feed uses the same GTINs your on-site schema uses. Consistency between feed and schema is what allows Google to build a high-confidence canonical entity. Discrepancies between the two create the kind of ambiguity that reduces your entity's authority in the graph.
Build your review ecosystem in a way that connects to the entity. Reviews marked up with Review schema on pages with valid GTIN data contribute directly to the Shopping Graph entity. Third-party platform reviews contribute too, less directly. Over time both matter, and neither replaces the other.
Do this consistently across drops and the advantage compounds. Your product entities get richer each cycle. Your review depth grows. Your authority over your own Product Cards strengthens. The brands that haven't done this work are competing for visibility in a graph they have limited presence in and that gap is considerably harder to close from behind than it is to build from the beginning.