Etsy's stock jumped 16% in a single day in September 2025, and the reason had nothing to do with earnings. OpenAI had just announced that ChatGPT users could buy directly from Etsy sellers inside a chat window, no browser tab required. That single product announcement told you everything about where product discovery is heading: away from ten blue links and into a conversation.
If you sell anything online, "ranking on Google" is no longer the whole job. A growing slice of your future customers are going to ask ChatGPT what to buy before they ever type a query into a search bar. This guide breaks down exactly how ChatGPT surfaces products, what's changed since the original Instant Checkout launch, and what a marketing or ecommerce team actually needs to do to show up in the results.
What ChatGPT Shopping actually is
"ChatGPT Shopping" isn't one single feature. It's a set of capabilities OpenAI has been stacking on top of each other since 2025:
- Product search and product cards. When someone asks ChatGPT for a recommendation ("best running shoes for flat feet under $150"), it can return visual product cards with images, prices, ratings and a link, pulled from real merchant data rather than generated from memory.
- Instant Checkout. Announced on September 29, 2025, this let US-based ChatGPT Plus, Pro and Free users complete a purchase without leaving the chat, starting with Etsy sellers.
- The Agentic Commerce Protocol (ACP). The open standard, built with Stripe, that handles the actual payment handoff between ChatGPT, the merchant, and the payment processor.
- Merchant partnerships. Direct integrations with retailers like Walmart, which let ChatGPT tap into a retailer's own catalog, loyalty program and payment rails instead of relying only on a generic feed.
I think it's worth separating these pieces because they're moving at different speeds. Product discovery (the search and cards part) is the layer OpenAI is investing in hardest right now. Checkout, on the other hand, has already gone through a public course correction, and that matters for how you should prioritize your effort.
How ChatGPT decides which products to show
When ChatGPT returns a shopping answer, it isn't improvising. According to OpenAI, product results in ChatGPT are organic and unsponsored, ranked on relevance to the query and on how complete and accurate the underlying product data is. There's no auction, no paid placement, at least not yet. That's a meaningfully different model than Google Shopping, where ad spend has always played a role in visibility.
What that means in practice: ranking runs primarily on three inputs.
- Feed quality. Merchants push structured product data (title, price, availability, images, shipping details, reviews) to OpenAI, and incomplete or stale data simply won't compete against a feed that's accurate down to the SKU.
- Freshness. Feeds can be refreshed as often as every 15 minutes, and OpenAI has said price and stock accuracy is a ranking input, not just a display detail. A product marked "in stock" that's actually sold out is the fastest way to get demoted.
- Whether you're the maker or primary seller. OpenAI has indicated it favors the actual brand or authorized seller over a reseller with a thinner listing, which is a departure from how marketplaces have historically ranked by price alone.
There's a fourth input that OpenAI hasn't fully detailed publicly but that shows up consistently in early testing: how much independent, citable content exists about a product elsewhere on the web. ChatGPT's shopping mode still draws on the same web-search and retrieval infrastructure behind its regular answers, the same fan-out queries it uses to research any other topic. A product with strong reviews, comparison coverage and a clean Google Shopping presence gives the model more signal to work with, even before your own feed enters the picture.
Instant Checkout: what changed, and why it matters for your priorities
Here's the part most "how to get on ChatGPT Shopping" guides skip, and it's the most important strategic detail in this whole space.
Instant Checkout launched with Etsy in September 2025 and expanded toward Shopify's more than one million merchants, including brands like Glossier, SKIMS, Spanx and Vuori. But by early 2026, OpenAI had quietly reworked the approach. Walmart's integration, announced in October 2025 and expanded through Walmart's Sparky assistant inside ChatGPT, runs on Walmart's own checkout and payment flow rather than the original one-size-fits-all Instant Checkout button. OpenAI itself acknowledged the shift, saying the initial version of Instant Checkout "did not offer the level of flexibility that we aspire to provide," and that the company is now letting merchants use their own checkout experiences while it concentrates on product discovery.
That's a meaningful signal if you're deciding where to put your team's time this quarter. Building a custom ACP checkout integration is a real engineering lift, and OpenAI's own messaging suggests discovery, getting your products found, described and shown accurately, is where the near-term leverage actually sits. Checkout flexibility matters more for large retailers with existing payment infrastructure than for a mid-size DTC brand still figuring out its feed.
Who's actually plugged in right now
| Partner | What's live | What it means for brands |
|---|---|---|
| Etsy | Instant Checkout for US sellers since September 2025 | Individual Etsy shops get purchase-ready product cards automatically |
| Shopify | Feed integration rolling out to 1M+ merchants | Any Shopify store can be feed-eligible without custom development |
| Walmart | Product feed plus Sparky assistant, using Walmart's own checkout | Walmart-listed brands inherit visibility without a direct OpenAI relationship |
| Independent/custom platforms | Manual feed submission via the merchant portal or the commerce API | Requires building your own feed pipeline and, for checkout, an ACP-compliant endpoint |
If your store runs on Shopify or you sell through Walmart, a chunk of this work is already being done for you at the platform level. If you run a custom stack or a headless commerce setup, you're the one who has to build and maintain the feed.
The practical checklist: getting your products cited in ChatGPT
This is the part that actually changes your results. None of it is exotic, most of it is just disciplined product data work that a lot of ecommerce teams have let slide for years.
- Apply through the merchant portal. Go to chatgpt.com/merchants and submit your company details, website, product categories and SKU count. OpenAI reviews and approves merchants before feed ingestion starts, and there's no cost to submit a feed or appear in results.
- Build a complete, accurate feed. OpenAI accepts CSV, TSV, XML or JSON, delivered either via SFTP or through its commerce API. Every field matters: titles that match how people actually search, real-time price and inventory, high-quality images, shipping and return details, and review counts where you have them.
- Refresh often, not once. Feeds can update as frequently as every 15 minutes. If your inventory turns over fast or your pricing changes with promotions, set up an automated pipeline rather than a manual upload someone remembers to run once a month.
- Fix the maker/reseller signal. If you're a brand selling through multiple resellers, make sure your own direct listing (or your most authoritative retail partner) is the version with the strongest, most complete data. A thin reseller listing competing against your own weak feed does you no favors.
- Don't neglect the open web. Since ChatGPT still pulls context from web search when it evaluates a shopping query, the same fundamentals from AEO and GEO work still apply: comparison content, genuine reviews, and clear structured product data (schema markup) on your own site all feed the model's broader understanding of your product before your feed even gets checked.
- Track what's actually showing up. You can't optimize what you can't see. Run your own prompts regularly to check whether your products appear, what competitors show up instead, and how your listings are described. A ChatGPT visibility tracker turns this from occasional spot-checking into an actual monitoring practice.
- Watch for checkout eligibility separately from discovery. Being discoverable in product cards and being eligible for Instant Checkout are two different states. Don't assume one guarantees the other, especially while OpenAI is still adjusting how checkout works platform by platform.
Common mistakes brands are making right now
A few patterns keep showing up when I look at how brands are (or aren't) approaching this.
The biggest one is treating ChatGPT Shopping like a one-time integration project instead of an ongoing data discipline. Submitting a feed once and walking away is the ecommerce equivalent of writing a blog post in 2015 and never updating it. Prices drift, stock changes, and a feed that was accurate in January can be quietly wrong by March.
The second mistake is ignoring the discovery layer while chasing checkout. Plenty of teams got excited about Instant Checkout headlines and started scoping ACP integrations before they'd even confirmed their basic product feed was complete. Given OpenAI's own acknowledgment that discovery is the current priority, that's backwards for most mid-size brands.
The third is assuming this only matters for huge retailers. It doesn't. A single-product DTC brand with a clean, accurate feed and strong independent reviews can outrank a much bigger competitor with sloppy data, because ranking here runs on completeness and relevance, not company size.
What this means for the ChatGPT shopping assistant experience
It's worth being precise about the ChatGPT shopping assistant itself, since the term gets used loosely. There isn't a separate app or a dedicated "shopping mode" toggle you switch on. It's the same ChatGPT conversation, and shopping behavior kicks in when the model detects commercial intent in what you type, a request for a recommendation, a price comparison, a "where can I buy" question. That's different from Amazon's Rufus or Perplexity's shopping features, which live in more clearly separated interfaces. For brands, this actually raises the bar: your product needs to compete for attention inside a general-purpose conversation, not just inside a dedicated shopping surface where people have already committed to browsing.
That also explains why so much of the optimization work overlaps with general AEO. If ChatGPT for shopping and ChatGPT for answering "what's the best noise-canceling headphone" pull from the same underlying retrieval and ranking logic, then a brand that's invisible in ChatGPT's regular answers is very unlikely to suddenly appear the moment someone adds "buy" to their prompt.
FAQ: ChatGPT shopping for brands and marketers
Does my brand need to be a huge retailer to show up? No. Ranking runs on data completeness and relevance, not company size or ad budget. A small brand with an accurate, well-structured feed can outrank a larger one with a messy catalog.
Do I have to build Instant Checkout to benefit from ChatGPT Shopping? No. Discovery (showing up in product cards and recommendations) and checkout are separate layers. OpenAI has said discovery is the current priority, and plenty of brands will get real value from being cited and recommended even without a checkout integration.
Is there a cost to get listed? OpenAI has stated there's no cost to submit a product feed or to appear in shopping results, since results are organic and unsponsored. That could change as the ad model matures, so it's worth checking current terms before assuming that stays true indefinitely.
How is this different from Google Shopping? Google Shopping listings have historically been influenced by ad spend through Google Merchant Center campaigns. ChatGPT's shopping results, at least as of today, are explicitly not ad-driven, which puts more relative weight on the quality of your product data and your presence elsewhere on the web.
What if I don't sell on Shopify, Etsy or through Walmart? You can still apply directly through the merchant portal at chatgpt.com/merchants and submit a feed via SFTP or the commerce API. It takes more manual setup, but it isn't limited to those platforms.
Where this is heading
ChatGPT Shopping is still young, and OpenAI has already shown it's willing to rebuild pieces of it in public when the first version doesn't work. That's actually useful information: it tells you this channel rewards brands that stay current rather than brands that set something up once and forget it. The teams that will own this category eighteen months from now are the ones treating their product feed with the same seriousness they treat their Google Shopping feed today, not the ones waiting for a stable, finished playbook that may never arrive.
If you're trying to figure out where your brand actually stands in ChatGPT, Perplexity and Google AI Overviews right now, that's exactly what we built AI Peekaboo to show you. Email us at filipe@aipeekaboo.com or book 30 minutes and we'll walk through your current visibility together.
