Optimising for ChatGPT shopping and optimising for Google Shopping may be the same job. A Semrush guide published November 4 found that 83% of products appearing in ChatGPT shopping carousels also rank among the top 40 Google Shopping results. A Semrush experiment cited in the guide found the top ChatGPT recommendation sat in the top three Google Shopping results 75% of the time. The overlap is so strong that the guide's core advice is familiar to any ecommerce SEO: accurate titles, GTIN and MPN identifiers, extra images, clean Merchant Center diagnostics and real review text on the page. Shopping in ChatGPT is live only in the US for now, but the playbook is already written — and it starts with Google.
The guide's most important detail is a technical one. ChatGPT reads raw server-sent HTML: OpenAI's crawlers do not execute JavaScript, so prices must live in the HTML itself, not in JavaScript-rendered content. A store whose prices only appear after scripts run is effectively invisible to ChatGPT's shopping layer, however good its products are. That is a rendering problem with a simple fix, but only if you know to look for it.
The same logic applies to the rest of the product page. Structured data, clean identifiers and real review text all help ChatGPT understand what a product is and whether it is worth recommending. None of this is new to ecommerce SEO — it is the same hygiene that Google Shopping rewards. The finding that 83% of ChatGPT carousel products already rank well on Google suggests ChatGPT is, for now, largely re-ranking the same well-optimised catalogue rather than discovering a different one.
OpenAI launched shopping inside ChatGPT in 2025, then added Instant Checkout through its Agentic Commerce Protocol, which let people buy without leaving the chat. In March 2026 it stepped back from Instant Checkout to focus on discovery instead — the browsing and comparing stage, rather than the transaction itself. That shift is good news for retailers: discovery is where product pages, reviews and rankings do the work, and it is where SEO skills transfer directly.
The guide's framing is that ChatGPT shopping is a discovery engine first. People ask for recommendations, compare options, and then click through to buy. The click still matters, and the product page still has to close the sale. For now the audience is US-only, which gives non-US retailers time to prepare — but the preparation is the same work they should already be doing for Google.
The 83% overlap is the most actionable number in the story. It means the work you already do for Google Shopping — feed quality, identifiers, reviews, page experience — is also the work that gets products into ChatGPT's carousels. There is no separate ChatGPT shopping optimisation to learn from scratch, at least not yet. The one exception is the JavaScript rule: check that prices render in raw HTML, because that is a failure mode Google's crawler would never expose.
For link builders, product pages that ChatGPT recommends earn a new kind of visibility: citation inside a conversation where the user is ready to buy. A link from a buying guide or a review roundup that ChatGPT trusts can put a product in front of shoppers who never touch Google. The overlap figure says those guides and pages are, in most cases, the same ones Google already rewards — so the strategy is to keep earning them, and to make sure the product pages they point to are readable by every crawler, not just Googlebot.
The overlap figure also hints at something bigger about how AI shopping will evolve. If ChatGPT's recommendations are largely drawn from the same well-optimised catalogue that Google Shopping ranks, then the two systems will converge on the same winners — the retailers with clean feeds, complete identifiers and genuine reviews. That convergence is good news for established ecommerce SEO, because it means the investment already made in feed quality pays twice. The retailers at risk are the ones whose product data lives only in JavaScript, whose identifiers are missing, or whose reviews are thin: they are invisible to one system already, and now they are invisible to two. Fixing the basics is not glamorous work, but the 83% figure says it is the work that decides who gets recommended.
Sources: Semrush Blog, November 4, 2026.
Related reading on this site: Google AI Mode Now Books Home Services — Angi, Thumbtack, Jobber and Housecall Pro, Google AI Mode Now Shows Up to Four Ads Under a Single Answer, Facebook Answers Drew an Estimated 40.6 Million Google Visits in October, Only 4.1% of ChatGPT's Cited Sources Rank in Google's Top 10, LoudFace Finds, Brand-Named Prompts Get Cited 41.8% of the Time. Non-Brand Prompts: 0%, and google-search-console-click-spikes-bot-traffic.
Linkbaits helps you create content that earns citations and backlinks. Start with the free tools and build your first linkable asset this week.
Start free trial