Seventy-seven percent of mobile searches now end without a click.1 The search landscape that most merchants built their product feeds for is shrinking, and a new one is forming around AI shopping engines that evaluate, recommend, and increasingly purchase products on behalf of shoppers.
Five AI engines are evaluating your product data right now, and each one is looking for something different. Perplexity, Amazon Rufus, ChatGPT, Microsoft Copilot, and Google Gemini have different data priorities, different sourcing methods, and different purchase behaviors. Most merchants send the same product data to all of them and assume it’s enough. A title optimized for Google Shopping relevance may perform well in Gemini and ChatGPT, but it won’t help Perplexity’s citation-driven recommendation engine or Rufus’s closed-loop product Q&A.
In our recent webinar on agentic commerce, Mark Batson, Head of GTM Technical Operations, and Stephanie Brown, Head of Product, profiled how each engine discovers and recommends products. This article distills those profiles into a practical guide, with a personality for each engine, the data each one prioritizes, and an optimization map you can hand to your feed team. Athos Commerce is an intelligent discovery and feed management platform that helps merchants optimize product data across all five engines from a single system.
Gartner predicts that by 2030, 20% of monetary transactions will be programmable, giving AI agents the economic agency to execute commerce autonomously.2 Brands that understand how each engine evaluates their catalog will capture visibility that compounds every week, while those treating all five as interchangeable will fall further behind.
Perplexity: The decision engine
Think of Perplexity as the shopper who reads every review, checks Reddit, watches three YouTube videos, and builds a comparison spreadsheet before buying anything.
Mark Batson called Perplexity “the thinking person’s shopping mall,” and the label fits. It prioritizes citations over ads, scanning Reddit threads, expert blog reviews, and YouTube content to synthesize a recommendation for complex, considered purchases. When a shopper asks “Which air purifier is best for a 500 sq ft room with a shedding dog?”, Perplexity doesn’t return a list of sponsored results. It builds an answer from third-party sources and tells the shopper exactly why it’s recommending one product over another.
That citation-driven approach changes what your product data needs to look like. Perplexity draws from external signals more than from your feed, so your optimization strategy needs to extend beyond feed management itself. Rich product descriptions tied to specific use cases, structured specs that answer detailed questions, and a strong presence on review sites and expert publications all contribute to whether Perplexity recommends you. The platform reaches approximately 45 million monthly active users and 170 million monthly visitors, with commerce activity accelerating.3 Perplexity shoppers spend 57% more per order than average, and AI-driven search traffic through the platform converts at 14.2% versus 2.8% from traditional search.4
Perplexity is not yet fully transactional for most merchants, though its “Buy with Pro” feature allows in-platform checkout for Pro subscribers in the US. The primary opportunity today is to build structured data markup and earn reviews on independent publications that Perplexity’s citation engine trusts, so your products are the ones it recommends when shoppers ask complex questions. A practical second step is joining the Perplexity Merchant Program, which feeds your product data directly into its ranking model and gives you more control over how your catalog appears in recommendations.
It’s also worth watching how Perplexity’s business model shapes its results. In February 2026, the company announced a pivot away from advertising to focus on shopping features and growing its Pro membership base.5 By choosing subscription and commerce revenue over ad revenue, Perplexity is betting that neutral, citation-driven results will keep users coming back. For merchants, that means organic authority and third-party validation carry more weight here than paid placement.
Amazon Rufus: The closed-loop giant
Rufus is the shopper who drives to the same store every week and never leaves without buying something. The entire journey takes place within Amazon’s ecosystem, from question to purchase.
Batson’s label for Rufus was “the closed-loop giant,” and the conversion data support it. Rufus bridges the intent shortfall that other engines don’t address. A shopper can ask, “Will these hiking boots fit a wide toe box?” and Rufus answers using product attributes, review sentiment, and Q&A data from the listing, then surfaces a purchase option without the shopper ever exiting Amazon. Customers who interact with Rufus during a shopping journey are 60% more likely to complete a purchase than those using traditional Amazon search, and the platform is projected to drive $10 billion in incremental sales.6
With 250 million monthly active users and interactions growing 210% year over year, Rufus has become the largest AI shopping assistant by user count.7 If you sell on Amazon, Rufus is already evaluating your listings, whether you’ve optimized for them or not.
The data requirements for Rufus differ from those of the other four engines. Highly specific attribute data matters most. Fit details, compatibility information, material specs, and technical measurements all feed the precise questions shoppers ask in conversation. A strong review corpus on Amazon itself is critical because Rufus synthesizes review sentiment to build its answers. Accurate inventory and fulfillment data close the loop by ensuring that what Rufus recommends is actually available for purchase.
ChatGPT: The solution search
ChatGPT is the visual browser that scrolls through product cards, compares options side by side, and asks, “Why this one over that one?” With 900 million weekly active users and a ranking as the fifth most visited website globally at 5.7 billion monthly visits, it represents the single largest standalone AI audience for product discovery.8
The “solution search” label reflects how shoppers actually use ChatGPT, which is to handle the research phase of a purchase. The interface displays visual product cards and carousels that function similarly to Google Shopping results. For Shopify and Etsy merchants in the US, ChatGPT already supports Instant Checkout directly within the chat window. hortly after our webinar, however, OpenAI discontinued Instant Checkout and now redirects shoppers to vendors’ own websites to complete purchases.9 The shift reinforces ChatGPT’s role as a discovery and recommendation engine rather than a closed-loop transaction platform. OpenAI has announced a dedicated feed specification and ecommerce merchant dashboard, though the dashboard remains invite-only for now.
The hidden connection that most merchants miss is how heavily ChatGPT relies on Google’s index. When ChatGPT can’t find product data via Bing or its own index, it triggers a browser agent to query Google Shopping as a fallback. Testing shows a 75% overlap between the top products ChatGPT recommends and the top three organic results on Google Shopping, with prices, retailers, and availability data often matching exactly.10 We explored this connection in depth in Your Google Shopping Feed Is Already Powering ChatGPT. For most merchants, the fastest path to ranking on ChatGPT is to maintain a high-quality Google Merchant Center feed.
Stephanie Brown
Optimization for ChatGPT starts with that Google Shopping foundation, meaning clean pricing, accurate availability, and structured attributes. Beyond the feed, review data matters because ChatGPT performs sentiment synthesis, reading the actual text of reviews to answer detailed shopper queries. If your reviews aren’t integrated into your feed, ChatGPT sources social proof from Reddit or third-party blogs instead, leaving you with no control over the narrative. If you’re not on Shopify, registering for the ChatGPT Merchant Dashboard beta should be a near-term priority.
Microsoft Copilot: The value engine
Copilot is the deal hunter who tracks prices for weeks and won’t buy until they’re certain they’re getting the best value. With 33 million standalone monthly active users, it reaches a segment that most merchants overlook.11
In the webinar, Batson labeled Copilot “the value engine,” and its design reflects that positioning. Built on the Bing index, it caters to budget-conscious and value-minded shoppers. Price Match alerts and 30-day and 90-day price trackers are built directly into the conversation, giving shoppers a reason to keep returning before they commit to a purchase. Bing shopping ads see a 45% higher click-through rate compared to other platforms, which suggests that shoppers arriving through this channel carry strong purchase intent.12
Copilot’s reach is smaller than that of ChatGPT or Gemini, but its distribution strategy gives it an outsized presence in a specific segment. Deep integration into Windows and the Edge browser means Copilot captures a desktop shopping demographic that skews less tech-literate and more value-oriented than the audiences using ChatGPT or Perplexity. Bing holds roughly 12% of the desktop search market share and 5% globally, but electronics and home goods dominate 45% of Bing shopping searches, so merchants in those categories should pay closer attention.13
Copilot is also moving toward closed-loop transactions. At NRF 2026, Microsoft launched Copilot Checkout, which lets shoppers complete purchases directly within the Copilot chat experience across Bing, MSN, and Edge.14 Shopify merchants are automatically enrolled with an option to opt out, and early data shows that shopping journeys involving Copilot are 194% more likely to result in a purchase than those without it.
Optimization for Copilot revolves around pricing signals. Competitive pricing, sale price annotations, and clear value propositions in your product highlights all influence whether Copilot surfaces your products to its deal-hunting audience. If you haven’t set up a Bing product catalog, you’re missing this engine entirely. And since Bing also serves as ChatGPT’s secondary data source, that oversight affects your visibility across two platforms.
Google Gemini: The omnichannel orchestrator
Gemini is the local concierge who wants it today, knows where the nearest store is, and will pick it up on the way home. Of all five engines, Gemini comes closest to collapsing the entire distance between a digital “want” and a physical “have.”
The “omnichannel orchestrator” label captures what makes Gemini different. It goes beyond product recommendations. Gemini can check local aisle-level stock at a nearby retailer and bridge online and offline commerce through a single conversational interface. The transaction layer behind this experience is Google’s Universal Commerce Protocol (UCP), launched at NRF in January 2026.[^18] UCP enables checkout directly within Gemini and other Google surfaces using Google Wallet, but it requires merchants to actively opt in through Merchant Center and complete a technical integration. This is not a tick-the-box setup. Merchants need to expose their ecommerce infrastructure through UCP’s checkout and identity-linking APIs, which means development work to connect inventory, pricing, fulfillment, and loyalty data. UCP also supports Identity Linking, which lets shoppers receive the same loyalty pricing and free-shipping thresholds they would get on a retailer’s own site across any UCP-integrated platform. For now, UCP is US-only and limited to eligible brands and retailers.
The reach numbers reflect Google’s distribution-first strategy. The Gemini app and web interface reached 750 million monthly active users in early 2026, up from 450 million a year earlier.15 But the real scale comes from AI Overviews embedded in standard Google search results, which drive over 2 billion monthly interactions and now appear in up to 60% of US queries.16 Most shoppers encounter Gemini’s AI without ever visiting the Gemini chatbot directly, and those AI Overviews have already driven a 58% drop in click-through rates for the number-one organic search result when they appear.17
The foundation for Gemini is the asset you likely already have, your Google Merchant Center feed. Every product recommendation Gemini makes draws from that data. Beyond the basics, local inventory data, shipping speed per product, and the product highlights and product details fields within Merchant Center deserve particular attention. Those fields are where Gemini looks for intent signals, the contextual metadata that helps it recommend your product for specific use cases rather than returning generic matches. Optimizing product highlights for intent (“best heavyweight breathable hoodie for cool-weather hiking”) rather than compliance (“men’s hoodie, charcoal grey, size L”) is the difference between being found and being recommended. If you have the development resources, opting into UCP through Merchant Center is a concrete next step that positions your catalog for in-conversation checkout as Google expands the program.
The optimization map: One product, five data needs
The most common mistake merchants make is optimizing their Google Shopping feed for compliance and assuming that coverage extends to every AI engine. It doesn’t. Each engine weighs different attributes, sources data through different channels, and evaluates trust through different signals. A single compliance-level feed leaves money on the table across at least three of the five engines profiled above.
The table below maps the key data requirements for each engine. Use it as a starting point for your feed team to identify where your current optimization falls short.
| Data requirement | Perplexity | Amazon Rufus | ChatGPT | Microsoft Copilot | Google Gemini |
|---|---|---|---|---|---|
| Title and description | Use-case driven, answers specific questions | Highly specific attributes (fit, compatibility, specs) | Google Shopping optimized with structured attributes | Value-focused with clear savings messaging | Intent-driven product highlights over keyword stuffing |
| Reviews and social proof | External citations (Reddit, expert blogs, YouTube) | Amazon review corpus and Q&A data | Feed-integrated review data for sentiment synthesis | Verified reviews with value/quality signals | Aggregate ratings and verified purchase flags |
| Pricing and deal signals | Less relevant (citation-driven) | Competitive within the Amazon marketplace | Clean, accurate pricing matching your site | Sale price annotations, price competitiveness | Sale price annotations trigger deal signals |
| Fulfillment and shipping | Not yet transactional for most | Accurate inventory and Prime eligibility | Redirects to vendor checkout; shipping data aids ranking | Shipping cost transparency for value shoppers | Local inventory, shipping speed, return policy; UCP checkout (opt-in, US only) |
| Feed connection | Indirect (web presence and citations) | Amazon Seller Central listings | Google Merchant Center feed (primary source) | Bing product feed | Google Merchant Center feed |
| Unique priority | Third-party authority and expert mentions | Conversational Q&A readiness | Site crawlability for non-Shopify merchants | 30/90-day price tracking competitiveness | Product highlights and details fields for intent |
Your Google Merchant Center feed remains the single most important asset because it directly powers Gemini and serves as ChatGPT’s primary source of product data. But treating it as the only feed you need means ignoring the specific requirements of Perplexity, Rufus, and Copilot, three engines that collectively represent over 325 million monthly users.
Where to start: Pick your two
Instead of trying to optimize for all five engines at once, pick the two with the most revenue exposure and start there.
For most merchants, that means Gemini and ChatGPT. You already have a Google Merchant Center feed, which powers both engines. The immediate work is upgrading that feed from compliance-level to intent-level by adding product highlights with use-case language, populating shipping speed and return policy data per product, and integrating your reviews into the feed. Shopify merchants already have access to ChatGPT’s Instant Checkout. Everyone else should register for the ChatGPT Merchant Dashboard beta and set up a draft ChatGPT feed aligned to OpenAI’s specification so you’re ready when your invitation arrives.
Your second move should be setting up a Bing product feed if you haven’t already. It powers Copilot directly and serves as ChatGPT’s secondary data source, so a single-feed setup improves your visibility across both engines.
If you sell on Amazon, Rufus is already live and evaluating your listings. Treat your Amazon product content with the same rigor you give your Google feed. Fill every attribute field, build your Q&A section, and invest in getting verified reviews.
Perplexity is the longer play. Focus on building your external citation footprint through product reviews on independent sites, expert mentions, and structured data markup that helps Perplexity’s citation engine find and trust your brand.
If you want to see how your product data looks through each engine’s eyes, Athos Commerce can run a feed audit that identifies exactly where your optimization falls short across all five platforms. The Athos feed management platform monitors product data quality, enriches catalog attributes for AI discovery, and syndicates optimized feeds across Google, Amazon, social commerce, and emerging AI shopping engines from a single dashboard, so your team manages one system instead of five. Watch the full webinar for live demos of the AI-powered feed audit and a complete overview of the agentic commerce landscape.
FAQs
What are AI shopping engines?
AI shopping engines are platforms that use artificial intelligence to research, compare, and recommend products on behalf of shoppers. The five major AI shopping engines driving product discovery in 2026 are Perplexity, Amazon Rufus, ChatGPT, Microsoft Copilot, and Google Gemini. Each evaluates product data differently, from Perplexity’s citation-driven recommendations to Rufus’s closed-loop purchasing within Amazon. Gartner predicts that by 2030, 20% of monetary transactions will be programmable, giving these AI agents the economic agency to execute purchases autonomously.
How does ChatGPT find products to recommend?
ChatGPT relies heavily on Google Shopping data to power its product recommendations. Research shows a 75% overlap between the top products ChatGPT recommends and the top three organic results on Google Shopping. When ChatGPT can’t find product data through Bing or its own index, it triggers a browser agent to query Google Shopping as a fallback. For most ecommerce merchants, maintaining a high-quality Google Merchant Center feed is the fastest path to ranking in ChatGPT’s product carousels.
Which AI shopping engine should ecommerce merchants optimize for first?
Start with Google Gemini and ChatGPT because your existing Google Merchant Center feed already powers both engines. Gemini reaches over 2 billion monthly interactions through AI Overviews in standard search results, and ChatGPT draws from Google Shopping data for 75% of its product recommendations. Your second priority should be setting up a Bing product feed, which powers Microsoft Copilot and serves as ChatGPT’s secondary data source.
What is the difference between AEO and GEO for ecommerce?
Answer Engine Optimization (AEO) is the practice of structuring product data so that AI assistants like Siri, Alexa, and Google Assistant can extract direct, confident answers to shopper questions. Generative Engine Optimization (GEO) goes further, optimizing content and product data so generative AI tools like ChatGPT, Gemini, and Perplexity surface your products in their recommendations and citations. Both disciplines share a common foundation in structured product data, but GEO also requires building external authority through reviews, expert mentions, and third-party citations.
How does Amazon Rufus affect product listings?
Amazon Rufus is an AI shopping assistant with 250 million monthly active users that evaluates product listings using attributes, review sentiment, and Q&A data. Customers who interact with Rufus during a shopping journey are 60% more likely to complete a purchase than those using traditional Amazon search. If you sell on Amazon, Rufus is already evaluating your listings. Optimization requires highly specific attribute data such as fit, compatibility, and material specs, along with a strong verified review corpus and accurate inventory data.
Why does my product need different data for each AI shopping engine?
Each AI shopping engine evaluates products using different criteria and data sources. Perplexity prioritizes third-party citations from Reddit and expert blogs. Amazon Rufus synthesizes review sentiment and Q&A data within its closed ecosystem. ChatGPT pulls from Google Shopping feeds. Microsoft Copilot tracks pricing competitiveness through Bing. Google Gemini evaluates intent signals in your product highlights and local inventory data. Sending the same compliance-level product data to all five engines means you’re optimized for one or two while being overlooked by the rest.
Sources & Further Reading
- The Digital Bloom. “2025 Organic Traffic Crisis Analysis Report.” The Digital Bloom, 2025. https://thedigitalbloom.com/learn/2025-organic-traffic-crisis-analysis-report/
- Gartner. “Top Predictions for IT Organizations and Users in 2026 and Beyond.” Gartner Press Release, October 21, 2025. https://www.gartner.com/en/newsroom/press-releases/2025-10-21-gartner-unveils-top-predictions-for-it-organizations-and-users-in-2026-and-beyond
- DemandSage. “Perplexity AI Statistics.” DemandSage, 2026. https://www.demandsage.com/perplexity-ai-statistics/
- Alhena AI. “Perplexity Shopping Merchants Setup Guide.” Alhena AI, 2026. https://alhena.ai/blog/perplexity-shopping-merchants-setup-guide/
- Search Engine Land. “Perplexity Stops Testing Advertising.” Search Engine Land, February 2026. https://searchengineland.com/perplexity-stops-testing-advertising-469452
- Fortune. “Amazon Rufus AI Shopping Assistant Chatbot: 10 Billion in Sales.” Fortune, November 2, 2025. https://fortune.com/2025/11/02/amazon-rufus-ai-shopping-assistant-chatbot-10-billion-sales-monetization/
- Nova Analytics. “Amazon Rufus Agentic Auto-Buy: 250 Million Users.” Nova Analytics, November 2025. https://www.novadata.io/resources/news/amazon-rufus-agentic-auto-buy-250-million-users
- TechCrunch. “ChatGPT Reaches 900M Weekly Active Users.” TechCrunch, February 27, 2026. https://techcrunch.com/2026/02/27/chatgpt-reaches-900m-weekly-active-users/
- TechCrunch. “OpenAI’s Plans to Make ChatGPT More Like Amazon Aren’t Going So Well.” TechCrunch, March 24, 2026. https://techcrunch.com/2026/03/24/openais-plans-to-make-chatgpt-more-like-amazon-arent-going-so-well
- Wells, Tom. “New Finding: ChatGPT Sources 83% of Its Carousel Products from Google Shopping via Shopping Query Fan-Outs.” Search Engine Land, March 5, 2026. https://searchengineland.com/new-finding-chatgpt-sources-83-of-its-carousel-products-from-google-shopping-via-shopping-query-fan-outs-470723
- AI Business Weekly. “Microsoft Copilot Statistics.” AI Business Weekly, 2025. https://aibusinessweekly.net/p/microsoft-copilot-statistics
- SeoProfy. “Bing Statistics.” SeoProfy, 2025-2026. https://seoprofy.com/blog/bing-statistics/
- SeoProfy. “Bing Statistics.” SeoProfy, 2025-2026. https://seoprofy.com/blog/bing-statistics/
- Microsoft Advertising. “Conversations That Convert: Copilot Checkout and Brand Agents.” Microsoft Advertising Blog, January 2026. https://about.ads.microsoft.com/en/blog/post/january-2026/conversations-that-convert-copilot-checkout-and-brand-agents
- TechCrunch. “Google’s Gemini App Has Surpassed 750M Monthly Active Users.” TechCrunch, February 4, 2026. https://techcrunch.com/2026/02/04/googles-gemini-app-has-surpassed-750m-monthly-active-users/
- AI Business Weekly. “Gemini AI Statistics.” AI Business Weekly, 2025-2026. https://aibusinessweekly.net/p/gemini-ai-statistics
- Ahrefs. “AI Overviews Reduce Clicks (Update).” Ahrefs Blog, February 2026. https://ahrefs.com/blog/ai-overviews-reduce-clicks-update/
