Something fundamental has changed in how people shop online. Instead of opening five browser tabs to compare products, an increasing number of consumers are typing a single question into ChatGPT—and trusting the AI to do the research for them. ChatGPT shopping research is no longer a niche behaviour: the platform now processes roughly 50 million shopping-related queries every day, and studies show that one in four users believe ChatGPT gives better product recommendations than Google. For ecommerce brands and digital marketers across Asia and beyond, this shift raises an urgent question: how do you make sure your products appear in those recommendations?

The good news is that the selection logic behind ChatGPT’s product carousel is more transparent than most brands assume. Research has decoded exactly where ChatGPT pulls its product data, which signals it weighs most heavily, and—crucially—what you can do to influence where your products appear. This guide breaks down the mechanics of ChatGPT shopping, explains the Google Shopping connection that sits at the heart of AI product discovery, and gives you a concrete, step-by-step optimisation playbook to put into action today.

Why ChatGPT Shopping Matters for Ecommerce Brands Right Now

The numbers describing AI-influenced shopping are no longer projections—they are present-day realities. Generative AI traffic to retail sites increased 4,700% year-over-year through mid-2025, making AI referral one of the fastest-growing acquisition channels in ecommerce. At the same time, research from G2 found that generative AI chatbots are now the single biggest influence over purchase shortlists, ranking ahead of review sites, vendor websites, and even salespeople. Sixty-one percent of consumers already use AI tools for shopping research, and once they arrive through an AI referral, they convert at a meaningfully higher rate than standard organic traffic.

OpenAI’s Shopping Research feature, launched in late 2025, accelerated this trend further. Powered by a specialised GPT-5 mini model trained specifically for shopping tasks, it asks clarifying questions about budget, preferences, and use case, then spends several minutes researching across the web before delivering a personalised buyer’s guide. The feature is available to all logged-in ChatGPT users on Free, Go, Plus, and Pro plans, meaning its reach spans hundreds of millions of people. For brands selling in categories like electronics, beauty, home and garden, or sports and outdoor—the categories OpenAI identifies as best suited to this kind of deep, specification-heavy research—the implication is direct: showing up in ChatGPT’s product carousel is now a meaningful commercial opportunity that sits alongside, not beneath, traditional SEO.

How ChatGPT Actually Selects Products to Recommend

ChatGPT does not recommend products from memory or a static database. When a user submits a shopping query, the platform runs what researchers call query fan-outs—a set of background web searches performed before any response is composed. For shopping queries, ChatGPT actually runs two distinct sets of fan-out searches: one contextual set that gathers information to write the conversational part of the answer, and a separate, shopping-specific set that retrieves the actual products displayed in the carousel. These two search types operate independently, which is why optimising for one does not automatically improve performance in the other.

According to OpenAI’s own documentation, the factors that determine which products surface include structured metadata from first-party and third-party providers (such as price and product description), the model’s prior knowledge, and contextual alignment with what the buyer actually wants. The system weighs query relevance highly—it interprets meaning and intent rather than matching keywords—so a product page that explicitly states the use case a shopper describes will consistently outperform a generic product page that claims to serve everyone. Crucially, there are no paid placements in the organic product carousel. Visibility must be earned through data quality, relevance, and authority signals.

The Google Shopping Connection: The Key Insight Every Brand Needs

The single most important piece of research to understand about ChatGPT shopping is this: a large-scale study analysing more than 43,000 ChatGPT carousel products across ten product verticals found that over 83% of products ChatGPT recommends in shopping carousels come directly from Google Shopping’s organic listings. The exact title match rate between ChatGPT carousel products and Google Shopping’s top 40 organic results was 45.8%; for Bing, by comparison, the exact match rate was less than 1%. This is not a coincidence—it reflects a deliberate architectural choice by OpenAI to leverage Google’s existing, deeply verified product data ecosystem.

What makes this finding so significant is that it simplifies the optimisation challenge considerably. ChatGPT’s product carousel is, at its core, a structured retrieval layer built on top of Google Shopping’s organic product index. The products users see when they ask ChatGPT for the best espresso machine under a certain budget are, in more than eight out of ten cases, products that already rank well in Google Shopping for the equivalent query. This means that for ecommerce brands, the highest-leverage investment is not building some entirely separate AI optimisation programme—it is improving your organic Google Shopping performance, which then automatically flows through to ChatGPT visibility. Your Google Merchant Center feed is the primary data source for both channels simultaneously.

This also explains why many brands find their ChatGPT visibility invisible despite having good website content: their product feeds are incomplete, their Google Shopping organic rank is weak, or their structured data is missing. Fixing these upstream issues is what unlocks the downstream AI shopping benefit. Our AI SEO team regularly encounters brands whose traditional SEO metrics look healthy while their AI product visibility is close to zero—precisely because AI shopping operates on a different, feed-and-schema-centred logic.

The ChatGPT Shopping Optimisation Playbook

Getting your products into ChatGPT’s shopping recommendations is not about gaming an algorithm—it is about making your product data so clear, complete, and trustworthy that AI can confidently surface your items to the right shoppers. The following steps address every layer of the process, from technical foundations to content strategy.

Step 1: Grant AI Crawler Access

Before any other optimisation matters, ChatGPT must be able to crawl your site. OpenAI operates two distinct crawlers: GPTBot, which collects training data, and OAI-SearchBot, which powers real-time search referrals and shopping recommendations. Blocking OAI-SearchBot in your robots.txt file will exclude your products from ChatGPT recommendations entirely, regardless of how good your product content is. Many brands inadvertently block both crawlers at once when trying to limit training data collection, which eliminates their AI shopping visibility as a side effect.

The recommended configuration is to explicitly allow OAI-SearchBot while blocking GPTBot if you prefer not to contribute to model training. Also ensure your product pages are included in your XML sitemap, use clean accessible URLs, and that your product data is not locked behind JavaScript that loads only after page hydration—AI crawlers typically do not wait for client-side JavaScript execution, meaning dynamically injected schema can go entirely unseen.

Step 2: Optimise Your Google Merchant Center Feed

Given that 83% of ChatGPT carousel products come from Google Shopping, your Google Merchant Center feed deserves dedicated attention as a standalone ranking objective—not just a supporting output of your paid campaigns. The goal is 95% or higher attribute completion across your highest-revenue products, with particular attention to the fields that drive organic Shopping rank and therefore flow directly into ChatGPT visibility.

• Product titles: Write descriptive, accurate titles that include brand, model, key specs, and how shoppers actually search—not internal product codes or vague marketing phrases.
• GTINs and MPNs: Complete GTIN and manufacturer part number data is essential for Google Shopping organic rank and is one of the fields OpenAI’s merchant programme checks for eligibility.
• Pricing accuracy: Keep prices current and consistent across your feed, your product pages, and your schema markup. Discrepancies between data sources are one of the fastest ways for your products to be deprioritised by AI recommendation engines.
• Availability signals: Always mark products accurately as in stock or out of stock. AI engines will stop recommending products they cannot confirm are purchasable.
• High-quality images: Include multiple, high-resolution product images. Image quality influences how products appear in the visual carousel format ChatGPT uses.
• Reviews and ratings: Star ratings and review counts influence organic Google Shopping rank, which directly impacts ChatGPT carousel inclusion. Review generation is not just a conversion tool—it is a discoverability signal.

For Shopify merchants, ChatGPT has already integrated directly with the Shopify Catalog, meaning product data flows through automatically. For WooCommerce, Magento, or custom-built stores, a dedicated feed management approach is necessary to keep your Merchant Center feed healthy. If you need support building or auditing an ecommerce site architecture that can feed these channels properly, our ecommerce web development team can help establish the right technical foundation.

Step 3: Implement Server-Rendered Product Schema (JSON-LD)

Structured data is the bridge between your product catalogue and the AI engines that recommend products in conversational responses. Research shows that 65% of pages cited by ChatGPT include structured data, and AI platforms actively use schema markup to identify, parse, and surface products inside shopping answers. When your schema is missing, malformed, or contradicts your visible page content, AI models treat the listing as low-confidence and skip it in favour of competitors with cleaner data.

The recommended format is JSON-LD, which holds nearly 90% market share among structured data formats because it is parsable as standalone JSON without full HTML traversal—exactly how AI crawlers prefer to process data. A critical implementation detail: your JSON-LD must be server-rendered in the static HTML head, not injected through client-side JavaScript after hydration. AI crawlers process pages at initial fetch and will not wait for dynamic content to load. At minimum, every product page should include the following fields:

• Name, brand, description: A clear, specific product name and a description that explains what the product is, who it is for, and what makes it different.
• Price and priceCurrency: Current, accurate pricing that matches what appears on the page.
• Availability: Use the appropriate schema.org availability value (InStock, OutOfStock, PreOrder) and keep it updated in real time.
• GTIN/SKU/MPN: Unique product identifiers that AI engines use to cross-reference your product against other data sources.
• AggregateRating: Star rating and review count, with values that exactly match the visible rating on your page—discrepancies trigger trust penalties.
• Images: Absolute image URLs pointing to high-resolution product photos on a CDN.
• Shipping and return details: OpenAI’s merchant programme requires shipping details and return policy markup before a product is eligible for conversational recommendations.

After implementation, validate every product URL using Google’s Rich Results Test and the Schema.org validator. Pay attention to warnings, not just errors—missing recommended fields like GTIN may not break your schema, but they push your listings below competitors whose data is more complete. Our SEO service team conducts structured data audits as part of broader AI search readiness assessments.

Step 4: Write AI-Ready Product Descriptions

ChatGPT’s Shopping Research model was trained with reinforcement learning specifically to understand product attributes, pricing contexts, specification comparisons, and shopping intent. This means your product copy needs to go beyond generic marketing language and speak directly to the kinds of questions buyers ask when they are ready to purchase. Vague descriptions that rely on brand recognition rather than specific details do not give AI enough to work with, and they will consistently lose to competitors whose pages explicitly answer the queries your customers are submitting.

Write descriptions that answer “who is this for?” and “what makes this different?” in concrete terms. For example, rather than describing a vacuum as “powerful and efficient,” specify the motor wattage, the floor types it is optimised for, the noise level in decibels, and the ideal apartment size. This kind of specificity serves two purposes: it improves semantic matching when ChatGPT interprets conversational queries, and it provides the factual data points that AI models extract when building product cards. Category-level buying guides and comparison content serve a similar function—a well-structured guide titled “How to Choose the Right Air Purifier for a Singapore HDB Flat” positions your brand as an authoritative source whenever AI systems answer related queries, and legitimately includes your products where they fit. This is exactly the kind of content marketing that now serves both traditional SEO and AI product discovery simultaneously.

Step 5: Build a Strong Third-Party Review Presence

ChatGPT does not rely solely on the reviews hosted on your own product pages. The system actively analyses sentiment, review volume, and off-site mentions to build product labels like “durable,” “quiet,” or “budget-friendly.” What people say about your product on independent review platforms, Reddit threads, and editorial comparison sites can carry more weight than your own product page copy. Detailed reviews that mention specific use cases, pros and cons, and real-world performance carry considerably more signal than a high volume of generic five-star ratings.

A practical review strategy for AI shopping visibility looks like this: prioritise getting your products reviewed on the high-authority sites that cover your category, run a product PR programme focused on editorial roundup features, and monitor overall product sentiment across third-party platforms rather than only tracking your on-site ratings. If your resellers or channel partners consistently outrank you in Google Shopping because they have more reviews on their product listings, they will win the ChatGPT carousel position too—even for your own branded products. Influencer partnerships can also accelerate earned media mentions across platforms that AI engines actively index. Our influencer marketing team and the StarScout AI platform can help identify and activate the right voices for your product category.

Step 6: Invest in GEO-Optimised Supporting Content

Generative Engine Optimisation (GEO) is the discipline of structuring your content and product data so that AI systems can retrieve, understand, and cite your brand when generating answers to buyer queries. It is distinct from—but complementary to—traditional SEO. Where SEO targets blue-link rankings, GEO targets citation in AI-generated responses. The practical differences matter: GEO requires use-case-specific product descriptions rather than keyword-optimised ones, machine-readable schema markup rather than just on-page text, and detailed third-party review content that AI systems can verify independently.

For ecommerce brands, the most effective GEO content investments include category-level buying guides that cover the criteria shoppers use when evaluating products like yours, FAQ schema on category and product pages that pre-answer the clarifying questions ChatGPT tends to ask, and comparison content that positions your products within the competitive set honestly and specifically. Research by Princeton found that adding citations, quotations, and statistics to content can lift visibility in generative engine responses by up to 40%. Every piece of supporting content should be structured for extraction by AI systems—short, clearly labelled sections, explicit use cases, and concrete data points—not written primarily for human narrative flow. Our broader AI marketing and AEO capabilities are built precisely around this emerging discipline.

AEO: Preparing for Agentic AI Commerce

The conversation around AI shopping is already moving beyond product recommendations into what researchers call agentic commerce—where AI systems do not just suggest products but actively complete purchases on behalf of users. OpenAI’s Instant Checkout feature and its partnerships with Shopify, Walmart, and other retailers are early signals of this direction. Google’s Universal Commerce Protocol is another infrastructure investment pointing toward a future where transactions happen directly inside the chat interface, without the user ever visiting a retailer’s website.

For brands preparing for this shift, Answer Engine Optimisation (AEO) adds a layer on top of GEO: while GEO gets you mentioned inside an AI answer, AEO gets you purchased by an AI agent. This requires machine-readable pricing exposed in real time, accurate and current inventory signals, structured return policies, and checkout flows that AI agents can navigate programmatically. The volume of agent-completed purchases is still small today, but the infrastructure is being built rapidly, and the brands that establish clean, agent-readable data architecture now will hold a significant first-mover advantage when agentic commerce scales. If your underlying website architecture needs updating to support these requirements, our website design and development team can assess and implement the necessary changes.

How to Measure and Track Your AI Shopping Visibility

One of the most significant challenges brands face with AI shopping is that traditional SEO dashboards do not measure it. Your keyword rankings and organic traffic reports can look stable while competitors are quietly capturing market share through ChatGPT, Perplexity, and Google AI Mode. Closing this measurement gap requires a different set of tracking methods and a new set of performance indicators to monitor alongside your existing metrics.

The most practical starting point is to manually test the queries your target customers would most likely submit to ChatGPT when shopping for products like yours. Run those prompts in ChatGPT, note whether your products appear, and document which competitors are showing up in your place. From there, dedicated AI visibility tools can automate this process at scale, tracking citation frequency, share of voice across AI engines, and the sentiment associated with your brand in AI-generated responses. Key indicators worth tracking include: carousel inclusion rate for your top 20 product SKUs, AI-referred traffic in your analytics (segmenting by source for ChatGPT, Perplexity, and AI Overviews), and the conversion rate of AI-referred sessions compared to organic search. Our AppearSearch AI tool provides brand and product visibility monitoring across AI search surfaces, giving you the concrete data needed to identify gaps and prioritise improvements. For local businesses that want to understand how they appear in AI-driven local discovery, LocalLead AI serves a similar function at the local level.

The principle to keep in mind is that AI shopping visibility and traditional SEO are now two sides of the same coin. Brands that align their SEO team, Merchant Center management, and content programme around a unified goal—rather than treating feed management, editorial content, and structured data as separate workstreams—are the ones building the compounding advantage that determines AI search share of voice over the next few years.

Conclusion

ChatGPT shopping research has moved from a curiosity to a commercially significant channel in a remarkably short time. The mechanics are now well-understood: ChatGPT runs shopping fan-out queries primarily through Google Shopping, which means your organic Google Shopping performance is the single most important lever for AI carousel inclusion. On top of that foundation, server-rendered JSON-LD product schema, clean AI crawler access, strong third-party review presence, and GEO-optimised content work together to increase the completeness and trustworthiness of the signals that AI systems use when deciding which products to recommend.

The brands that move on this now—auditing their Merchant Center feeds, fixing their structured data, and building the right supporting content—will accumulate authority signals that make future AI citations progressively more likely. Those that wait will find the gap harder to close with each passing month. Whether you are building an ecommerce presence from scratch or optimising an established catalogue, the core message is the same: be machine-readable, be accurate, and be found.