42% of ChatGPT Answers Now Include Shopping Cards. Other AI Platforms Almost Never Do.

Sill queries four AI platforms daily through their actual chat interfaces: ChatGPT, Gemini, Google AI Overviews, and Perplexity. Each platform receives the same purchase-intent queries for each brand. When we analyzed the structured response data, one platform stood apart.

Over 40% of ChatGPT responses to purchase-intent queries included structured shopping cards: product listings with titles, prices, ratings, merchant offers, availability status, and direct buy links. These are not inline text mentions. They are interactive commerce modules embedded directly in the AI answer, each product linking to a retailer with ?utm_source=chatgpt.com appended to the URL.

The same purchase-intent queries sent to Perplexity, Gemini, and Google AI Overviews almost never triggered shopping cards. All three platforms support the feature in their APIs, but ChatGPT activates it at an order of magnitude higher rate than any other platform.

ChatGPT's shopping cards are not simple product mentions. They are structured commerce modules with the following characteristics:

• ~8 products per response on average, up to 16 in a single answer. Each product includes a title, price, and direct buy link.
• Multi-merchant comparison built in. Nearly every product shows offers from two or three retailers, letting buyers compare prices within the AI answer.
• Star ratings on 84% of products. Aggregated review scores are embedded directly in the card.
• UTM tracking on every link. Product URLs carry ?utm_source=chatgpt.com, meaning OpenAI is measuring downstream conversion from these placements.
• Availability and delivery info. Cards show "In stock online" and estimated delivery windows — the same metadata you see on Amazon or Google Shopping.

Each product listing includes a title, price, direct buy link, availability status ("In stock online, Free delivery"), and typically two to three offers from different retailers for the same item. 84.2% include aggregated star ratings. Nearly every product URL carries ChatGPT's UTM tracking parameter, meaning OpenAI is measuring downstream conversion from these placements.

This is comparison shopping built into the answer. When a buyer asks ChatGPT "best gaming headset under $100," they do not get a list of brand names. They get a product card with a Razer BlackShark V3 at $99.99 from razer.com, a SteelSeries Apex from Walmart at $31.99, and a HyperX Cloud from Best Buy at $49.99 — each with ratings, stock status, and a one-click buy link. The recommendation and the transaction are collapsing into a single interface.

Shopping cards surface products from a mix of major retailers and brand-direct stores. The top positions go to the same retailers that dominate Google Shopping: Best Buy, Walmart, Target, GameStop. But unlike Google Shopping, where brand.com listings typically appear below paid retailer ads, ChatGPT's shopping cards give DTC storefronts meaningful share of shelf. Brand-direct stores appear alongside major retailers with equal visual weight.

For consumer brands that sell direct, this is a structural advantage over traditional paid shopping channels. A buyer asking ChatGPT "best gaming headset under $100" might see a product card with offers from both the brand's own store and two retailers, each with price, stock status, and shipping info. The brand-direct link gets the same prominence as Walmart or Best Buy.

Shopping cards are not uniformly distributed. ChatGPT appears to activate them when it detects a query for a purchasable physical product — the kind of item a buyer could add to a cart. Queries about consumer electronics, fitness devices, and sporting goods consistently trigger shopping cards. Queries about SaaS, professional services, and B2B solutions do not.

This creates a two-tier AI recommendation system. For physical goods, ChatGPT is becoming a shopping engine: product comparisons with prices and buy buttons. For services and B2B, it remains a text-based recommendation engine. The implications for Share of Voice measurement are significant: in product categories, the "answer" now includes price competition alongside brand recommendations.

This is not a capability gap. Perplexity's API documentation explicitly describes shopping card extraction that triggers "when shopping intent is detected." Google AI Mode's API returns a shopping_cards field "when shopping intent is detected." Both platforms have built the infrastructure. But for the same purchase-intent queries that trigger ChatGPT's shopping cards over 40% of the time, these platforms rarely activate commerce features.

The divergence is a strategic choice, not a technical limitation. A Semrush study of 43,000+ ChatGPT carousel products found that 83% were sourced from Google Shopping's top 40 organic results. ChatGPT is effectively repackaging Google's product data into a conversational shopping experience. OpenAI launched ads on ChatGPT in February 2026 at approximately $60 CPM with a $200K minimum buy-in, and built Instant Checkout with Stripe — shopping cards are the infrastructure for commerce monetization. OpenAI is building a shopping engine inside a chatbot. Google and Perplexity, at least for now, are keeping their chat interfaces separate from their commerce layers.

The implication: ChatGPT is not just an AI recommendation platform anymore. For product categories, it is a marketplace. The brands and retailers that appear in shopping cards occupy a qualitatively different position than brands merely mentioned in text. A text mention says "consider this brand." A shopping card says "buy this product here for $99.99."

OpenAI's Agentic Commerce Protocol, launched in partnership with Stripe, is designed to collapse product discovery, comparison, and checkout into a single AI interface. Target, DoorDash, and Instacart have already built ChatGPT integrations. The shopping cards we are observing are the visible surface of this infrastructure.

For consumer product brands, the implications are immediate:

1. Product data accuracy matters. Shopping cards pull prices, availability, and ratings from retailer feeds. If your product data is stale, your listing will show incorrect prices or "out of stock" while competitors show live inventory.
2. DTC presence is viable. Unlike Google Shopping where brand.com listings are buried below paid retailer ads, ChatGPT shopping cards give brand-direct stores meaningful representation. Logitech, Razer, HyperX, and Corsair all appear alongside Best Buy and Walmart.
3. AI visibility now includes price competition. Traditional Share of Voice measures whether your brand is mentioned. On ChatGPT, the answer now also includes at what price, from which retailer, and whether you are the cheapest option. Monitoring must capture commerce metadata, not just text mentions.
4. Platform-specific strategy is essential. The divergence between platforms extends beyond SOV scores and sentiment. ChatGPT is a shopping engine. Perplexity is a research engine. Google AI Overviews is a search engine. Each demands different optimization.

For B2B brands, the timeline is longer. Shopping cards currently activate only for physical products. But the infrastructure exists on Perplexity and Google AI Mode, waiting to be activated. When B2B commerce cards arrive, the brands that are already visible in text recommendations will have a head start.

Shopping cards are the most visible form of AI commerce, but they are not where most AI-driven conversion happens. Measure Protocol analyzed 142,965 ChatGPT sessions and found that 21.6% of all ChatGPT interactions show commercial intent — far more than the ~2% that result in explicit shopping queries. For SaaS, professional services, and B2B, the conversion path never touches a product card. A buyer asks ChatGPT which project management tool fits a distributed team, reads a text-only recommendation, and navigates directly to the vendor's website. No card. No price comparison. But a real conversion influenced by AI.

The conversion data supports this. Superprompt's analysis of 12.3 million visits across 347 businesses (including SaaS, ecommerce, and professional services) found AI-referred visitors convert at 14.2%, compared to 2.8% from Google — a 5x difference. Adobe Analytics found AI-referred shoppers are 33% less likely to bounce and convert 31% higher than non-AI traffic. These conversions are happening across all categories, not just the ones with shopping cards.

Shopping cards matter most for DTC and retail brands selling physical products. For everyone else, the text recommendation is the conversion event. Either way, what the AI says about your brand — and whether it says anything at all — is what determines whether you are in the consideration set. The monitoring challenge is the same regardless of whether the buyer clicks a product card or types your URL after reading a text recommendation.

Sill's chat-layer monitoring captures the full response from each platform, including structured data that API-based tools miss entirely. Shopping cards, product offers, merchant domains, prices, ratings, and availability status are all extracted and stored alongside SOV analysis, sentiment, and citation data.

This is why monitoring the actual chat interface matters. An API call to GPT-4 does not return shopping cards. It returns text. The commerce layer is a chat-only feature, invisible to any monitoring tool that queries the API instead of the interface. If your monitoring tool shows you text-only SOV data for ChatGPT, it is missing 42% of the buyer experience for product queries.

The gap between what an API returns and what a buyer sees is widening. Shopping cards are the latest example, but they will not be the last. As AI platforms build commerce infrastructure, the tools that monitor the surface buyers actually interact with will produce the only measurements that matter.