By
Austin Coker
Reading time: 15 min
AI shopping agents don’t share a ranking system. Each one pulls from different product data, weighs different signals and sends the buyer to a different checkout. OpenAI and Google have published a fair amount about how they pick products. Meta and Amazon haven’t published a ranking method, so for those two you work from what each company has said about its data sources.
In this article, you’ll see how ChatGPT, Google AI Mode, Meta Muse and Alexa for Shopping each decide what to recommend, how they differ, and what you can do as a Shopify or Amazon seller to show up more often. Everything here comes from each company’s own documentation and announcements, linked inline.
AI shopping agents are AI assistants that take a shopping request in plain language, research products for the shopper, and either hand them off to a store or complete the purchase on their behalf. They sit somewhere between a search engine and a personal shopper.
The big shift for merchants is where the decision happens. In classic search, a shopper clicks through ten links and compares on your site. With an agent, the comparison happens inside the agent, and your product either makes the shortlist or it doesn’t. For the wider category, including on-site assistants, see our guide to AI shopping assistants.
All four agents covered here are live for US shoppers. They differ most in what data they trust, how much control you get over that data, and whether money can buy a placement.
OpenAI is more open about this than anyone else on the list. According to its help page on shopping in ChatGPT, a product appears in the carousel when ChatGPT perceives it as relevant to what the shopper wants. It looks at the query plus context such as Memory and Custom instructions, so two shoppers typing the same prompt can see different products.
The model considers structured metadata from first-party and third-party providers, like price and description, along with other third-party content. OpenAI’s example factors are available options, price, reviews and ease of use. If a shopper gives a budget, ChatGPT shifts its focus toward price.
A few details matter more than they first appear. ChatGPT may write its own simplified titles and descriptions for your products. Labels like “Budget-friendly” or “Most popular” are generated by the model and aren’t verified. Review summaries come from reviews on public websites, and OpenAI doesn’t verify those either. Your product page is not the only thing shaping how you’re described. What other sites say about you counts too.
When ChatGPT lists merchants for a product, it ranks them by availability, price, quality, and whether the merchant is the maker or primary seller. That last factor can favor brands that sell direct over resellers. Prices come from third-party providers and can lag. OpenAI also says product results are selected independently, are not ads and aren’t influenced by partnerships. Ads run separately, which we cover in our ChatGPT Ads guide.
On the data side, OpenAI’s merchant page says Shopify and Etsy catalogs are already integrated, with no setup needed. Other stores can apply to share a product feed. You don’t need a feed if ChatGPT already crawls your site, but OpenAI says feeds give you more control and that “richer data helps your products surface with more accurate information.”
Checkout has changed. OpenAI says it is “moving away from a standalone Instant Checkout experience,” and purchases are completed on merchant-owned websites or apps. In its March 24 announcement, OpenAI said the first version of Instant Checkout “did not offer the level of flexibility” it wanted, and extended its Agentic Commerce Protocol (ACP) to product discovery instead. Retailers including Target, Sephora, Nordstrom, Best Buy and Wayfair share feeds through it. Per OpenAI’s developer docs, feed onboarding is open to approved partners, the full feed is usually sent once a day with updates through the API, and promotions can only be sent through the API. OpenAI says there are no fees on purchases that start in ChatGPT.
What moves the needle:
Google AI Mode runs on the Shopping Graph. Google says the graph holds more than 50 billion product listings, with 2 billion updated every hour. AI Mode answers show price, reviews and inventory, and can build comparison tables with insights pulled from reviews.
The selection starts with query fan-out. Google explained at I/O that AI Mode runs several searches at once to work out what makes a product good for a specific need, then uses those criteria to suggest products. Ask for a travel bag for a rainy trip, and AI Mode first works out that waterproofing and easy-access pockets matter, then looks for listings that have them. We break this process down in our query fan-out guide.
The Shopping Graph builds listings from Merchant Center and from what retailers and brands post across the web. It also pulls images, descriptions, reviews and YouTube videos to understand a product’s characteristics. A missing attribute in your feed means Google has less to match against the criteria it generated.
Google’s official position is short. Its Search Central page on AI features and your website says there are no additional requirements or special optimizations for AI Mode. SEO fundamentals apply, structured data should match the visible text, and your Merchant Center and Business Profile should be up to date. You don’t need AI text files or special schema. AI Mode traffic is counted in the Search Console Performance report under “Web.”
Google is also giving merchants a view into AI results. AI performance insights in Merchant Center, announced May 27, show your share of voice in AI Mode, AI Overviews and the Gemini app against similar brands. They add product term insights, product attribute insights and an attribute completeness score that flags products missing key attributes. Google says the report is rolling out in the US, Canada, Australia, India and New Zealand.
For checkout, Google’s agentic checkout lets a shopper track a price and have Google buy the item on the merchant’s site with Google Pay after they confirm, for eligible US merchants. Separately, a checkout button powered by the Universal Commerce Protocol can appear on eligible listings in AI Mode and Gemini. It’s limited to select merchants who apply, it relies on the native_commerce product attribute, and you stay the seller of record.
Paid placement is part of the picture. Google started testing ads in AI Mode, and Performance Max, Shopping and broad match Search campaigns, including AI Max, are eligible to appear in AI Overviews and AI Mode. For the organic side, see our Google AI Mode guide.
What moves the needle:
Meta introduced Muse on September 8 as a personal AI agent. It runs on a secure virtual machine with its own browser, so it can open sites and fill out forms. It remembers preferences and checks with the person before making a purchase. At launch, checkout ran through Link by Stripe, which generates a one-time-use card. Muse is rolling out in the US on iOS, Android and muse.ai, and people can talk to it in the Muse app or in WhatsApp.
Meta hasn’t published how Muse ranks products. What’s public is where it looks. In its Connect recap, Meta said Muse launched with access to the entire Shopify catalog and is adding Walmart, Best Buy, Gap, Sephora, Ulta, Wayfair and other retailers, plus Shop Pay and PayPal for payments. CNBC reported that Muse browses retail sites to find items that fit the request, suggests listings, and shows an approval card before it buys. CNBC also reported that PayPal and some connectors were announced but not yet live.
For Shopify merchants, the practical details are on Shopify’s Selling on Meta help page. Product discovery runs on Shopify Catalog, and direct checkout is on by default for eligible stores. To be eligible, you need to sell to customers in the US, Canada, Mexico, Brazil, Japan or Singapore, have Catalog-eligible products, and have your terms of service, privacy policy and return policy filled in. Shopify asks that legal disclosures sit within the first 6,000 characters of your product description. Subscriptions, bundles, customizable products and B2B-only products aren’t supported in direct checkout, and Google Analytics and custom pixels don’t fire on those orders. You manage it all under Sales channels > Agentic > Meta AI and Muse.
Fees are worth watching. Shopify says there are no fees for Meta’s direct checkout beyond standard payment processing. CNBC reported that Mark Zuckerberg said Meta will take a small fee on transactions. Neither source spells out who carries that fee, so check your own order data once Muse sales come in.
We cover setup step by step in our Meta Muse for Shopify guide.
What moves the needle:
On May 13, Amazon renamed Rufus to Alexa for Shopping. Amazon says Rufus was trained on its product catalog, customer reviews, community Q&As and information from across the web. On product pages, it answered questions from listing details, reviews and community Q&As.
Alexa for Shopping goes further. According to Amazon’s announcement, it combines Rufus’s product knowledge with the personal context of Alexa+: preferences, shopping history and conversations across Amazon and Alexa. Shoppers can ask questions in the main Amazon search bar. It shows AI overviews in search results and on product pages, builds side-by-side comparisons of features, prices and reviews, and shows up to a year of price history. It can run scheduled actions, and through Buy for Me it can buy eligible products from other stores on the shopper’s behalf. It’s free for all US customers, with no Prime membership required. Amazon says Rufus helped more than 300 million customers in 2025.
Amazon hasn’t published a ranking formula for Alexa for Shopping. Based on what it draws from, your listing content, your reviews and your Q&A coverage are the inputs you control. Because it uses shopping history, results vary by shopper, much like ChatGPT with Memory.
Ads are already inside the experience. In a June 11 Amazon Ads post, Amazon said sponsored ads appear in Alexa for Shopping and existing campaigns are automatically eligible. Sponsored Products and Sponsored Brands prompts open product conversations with the assistant. Amazon says nearly 20% of shoppers who interact with a prompt continue the conversation about that brand, and that adding prompts to a Sponsored Brands ad drives a 6% increase in conversions. Those are Amazon’s own figures. Our earlier Amazon Rufus guide still covers the listing fundamentals.
What moves the needle:
We’ll run your buying prompts across ChatGPT, AI Mode, Muse and Alexa for Shopping and show you who’s winning them.
ChatGPT recommends. Muse acts. ChatGPT shows a carousel and sends the shopper to your site to buy. Muse uses its own browser to find options and completes the purchase once the shopper approves.
Transparency is the other big gap. OpenAI has published the factors it weighs, how it ranks merchants, and that partnerships don’t influence results. Meta hasn’t published comparable detail for Muse.
Both lean on Shopify Catalog, so an eligible Shopify store is in both by default. The economics may differ. OpenAI says there are no fees on purchases that start in ChatGPT, while Meta has said it will take a small fee on Muse transactions.
Attribution is harder with Muse. Shopify notes that client-side pixels don’t fire on Meta direct checkout orders, so you’ll rely on Shopify’s channel attribution to see what Muse drives.
Google reports on AI Mode, AI Overviews and the Gemini app together in Merchant Center, and its UCP checkout covers both AI Mode and Gemini. In practice, the same feed work serves Google’s AI surfaces as a group.
ChatGPT assembles its picture from many providers, feeds and public reviews. Google starts from one index it refreshes constantly. With Google, Merchant Center is your control panel. With ChatGPT, your feed helps, but third-party content and reviews carry real weight in how you’re described.
Ads separate them sharply. In Google, your Shopping and Performance Max campaigns can appear inside AI Mode. In ChatGPT, OpenAI says product results are not ads. Google also offers more ways to check out inside the experience, while ChatGPT now hands off to your store.
For how this work compares with classic search, see GEO vs SEO.
Every agent rewards complete, accurate product data. Whether it comes from Shopify Catalog, Merchant Center, a direct feed or an Amazon listing, gaps in your data become gaps in your visibility.
Reviews matter everywhere. ChatGPT summarizes public reviews, Google builds comparison insights from them, and Amazon’s assistant answers questions from them.
Personal context is rising. ChatGPT uses Memory and Custom instructions, Muse remembers preferences, and Alexa for Shopping uses shopping history. There’s no single rank to track anymore. You track how often you show up across many realistic prompts.
Shopify is the common thread. Through Shopify’s agentic storefronts, eligible products are available to ChatGPT, Google AI Mode and Gemini, Microsoft Copilot and Meta, and the feature is on by default for eligible stores. For Google, products flow through the Google & YouTube sales channel or Merchant Center.
Open Sales channels > Agentic in Shopify and check which AI channels are active and whether your products are Catalog-eligible. Fill in your terms, privacy and return policies. If you’re not on Shopify or Etsy, apply to share a product feed with OpenAI and keep Merchant Center healthy for Google.
Treat Merchant Center attributes, Shopify product fields and Amazon listing fields as ranking inputs. Once AI performance insights reaches your Merchant Center account, use the attribute completeness score to find the gaps.
Think about the criteria an agent would generate for your category, such as materials, sizing, use cases or ingredients. Answer them plainly on the page, and keep structured data matched to the visible text.
Collect reviews on your site, on Amazon and on public sites ChatGPT can read. Answer community Q&As on your Amazon listings. Don’t buy or fake reviews; the goal is evidence a model can verify.
List the prompts real customers use and run them across all four agents on a regular schedule. Record whether you appear, which product shows up and which competitors win. Use Merchant Center’s share of voice for Google. This is how we found the gap for Buoy, which became ChatGPT’s #1 recommendation for “best electrolytes with no sweeteners” in three weeks.
Google and Amazon let your existing campaigns reach their agents. ChatGPT keeps organic product results and ads separate, and Muse has no announced sponsored placements. Budget accordingly, and don’t expect ad spend to lift organic recommendations.
AI shopping agents are one part of generative engine optimization. The same product data, content and third-party coverage that win shopping prompts also get your brand named when buyers ask which brands to trust.
At 95 Projects, we help Shopify and DTC brands show up when shoppers ask AI agents what to buy. You work with senior people, not juniors learning on your account. Our GEO for Shopify work covers product data, on-page content, reviews and third-party coverage, with prompt tracking across ChatGPT, Google AI Mode, Muse and Alexa for Shopping, measured in revenue. If you want to see where your products stand today, book a call and we’ll walk through it with you. You can also read more about our AI search services.
Austin Coker is the CEO and founder of 95 Projects. He has worked in SEO since 2020, when he built and sold an ecommerce brand that grew mainly through search. He founded 95 Projects in 2022 to help SaaS, ecommerce, and B2B service companies show up wherever their buyers search, from Google to ChatGPT and AI Overviews.
Have questions about working with us? Book a 30-minute strategy call to discuss your goals and see if we’re a good fit.
No-pressure conversation
Clear next steps (if we’re a fit)
Talk directly to an expert
It depends on where your customers buy. If you sell heavily on Amazon, Alexa for Shopping sits in the search bar your buyers already use. For DTC Shopify brands, ChatGPT and Google AI Mode are the bigger discovery channels, and Muse is worth checking early since eligible Shopify stores are included by default.
ChatGPT recommends products and sends shoppers to the merchant’s site to buy. Muse is an agent that browses, prepares the order and completes the purchase after the shopper approves it. OpenAI has published how ChatGPT picks products; Meta hasn’t published how Muse ranks them.
No. OpenAI says you don’t need a feed if ChatGPT already crawls your site, and Shopify and Etsy catalogs are already integrated. A feed still gives you more control over how your products appear and keeps prices and availability more accurate.
No. OpenAI says product results are selected independently, aren’t ads and aren’t influenced by partnerships. ChatGPT does show ads, but they appear separately from product results.
No. Google says there are no additional requirements or special optimizations for AI Mode. SEO fundamentals, structured data that matches the page and an up-to-date Merchant Center account are what Google recommends.
Not under that name. Amazon renamed Rufus to Alexa for Shopping on May 13, 2026. It combines Rufus’s product knowledge with Alexa+ personal context and is free for all US Amazon customers.
It varies. OpenAI says there are no fees on purchases that start in ChatGPT, and Shopify says there are no fees for Meta’s direct checkout beyond payment processing. CNBC reported that Meta will take a small fee on Muse transactions, so watch your order data.
