Meta Joins the AI Shopping Race. That Makes Five Platforms Retailers Need to Feed.

Bloomberg reported on March 3 that Meta is testing an AI-powered shopping research tool inside Meta AI. The feature shows product carousels with pricing, brand information, and links to merchant websites. No native checkout - just referral links, similar to how Google Shopping has worked for years. Forrester analyst Sucharita Kodali called it "a copycat move," and she's right. But that doesn't make it irrelevant.

Count the major AI shopping surfaces retailers now need to think about: ChatGPT, Google AI Mode, Perplexity, Amazon Rufus, and now Meta AI. That's five platforms where consumers can discover products through conversational AI, each pulling product data in different ways, each with its own ranking logic. For most retail teams, keeping one Google Shopping feed accurate is already a struggle. This just got five times harder.

How does Meta AI shopping work?

Meta's AI shopping tool, currently in testing within the Meta AI assistant, presents product recommendations as visual carousels showing price, brand name, and product images. Users research products through natural language queries and receive curated results. Unlike OpenAI's now-abandoned Instant Checkout, Meta opted for a referral model - clicking a product sends the user to the merchant's website to complete the purchase. This approach avoids the inventory and fulfillment complexity that undermined ChatGPT's native checkout attempt.

Meta choosing referral over native checkout looks smart in hindsight - or maybe just in light of OpenAI's retreat this same week. They avoid becoming the merchant of record, dodge the inventory synchronization nightmare, and still capture the valuable discovery moment. The ad revenue model writes itself: merchants pay for placement in AI-generated product recommendations the same way they pay for placement in the Facebook and Instagram feeds today.

With 3.3 billion monthly active users across Meta's apps, even a small percentage using the AI shopping tool represents enormous volume. We're already seeing 41% of consumers using AI for product discovery, with a third having ditched traditional search entirely. The question is whether Meta AI can surface genuinely useful product recommendations or whether it becomes another ad-cluttered feed that consumers learn to ignore.

How big is the AI shopping assistant market?

According to Grand View Research, the global AI shopping assistant market was valued at $3.36 billion in 2024 and is projected to reach $28.54 billion by 2033, growing at a 27% compound annual growth rate. North America accounts for approximately 40% of the worldwide market. Growth is driven by conversational AI improvements, increasing consumer comfort with AI-mediated product discovery, and platform competition among major tech companies to capture shopping intent.

That growth trajectory explains why every major tech platform is rushing to add shopping features to their AI products. Amazon has Rufus and its controversial Buy for Me feature. Google has AI Mode with shopping integration. Perplexity launched shopping months ago. OpenAI built (and then killed) native checkout. Now Meta. Each platform sees the same $28 billion opportunity and wants a piece of it.

The 27% CAGR also means this market is moving faster than most retail operations teams can adapt. A retailer that takes six months to optimize their product data for one new AI surface will find two more surfaces have launched by the time they finish.

How many AI shopping platforms do retailers need to support?

As of March 2026, retailers face at least five major AI-powered shopping discovery platforms: ChatGPT (OpenAI), Google AI Mode, Perplexity, Amazon Rufus, and Meta AI. Each platform ingests product data differently - some crawl merchant websites, some pull from structured feeds, some use partnership APIs. Additionally, Klarna (powered by OpenAI), Shopify's AI features, and eBay's AI tools represent secondary surfaces. There is no unified standard for AI product discovery optimization, making multi-platform presence operationally complex.

This is the part that keeps me up at night from an industry perspective. SEO took a decade to mature, and retailers eventually figured out how to optimize for Google. But Google was basically the only search engine that mattered. Now there are five AI surfaces that matter, and each one works differently. ChatGPT relies on partnerships and web crawling. Google AI Mode has direct access to the Shopping Graph. Perplexity crawls the web independently. Amazon Rufus works from Amazon's own catalog. Meta AI will likely leverage its commerce partnerships and advertiser data.

A retailer with a messy product catalog - inconsistent titles, missing attributes, outdated pricing - was already losing visibility on Google Shopping. Now that same mess is replicated across five AI surfaces, each amplifying the data quality problems in different ways. An AI that confidently recommends a competitor's product because your listing had incomplete specs is worse than not appearing at all. This is the branded vs. non-branded visibility gap multiplied across five platforms.

What should retailers prioritize for multi-platform AI discovery?

Retailers should focus on structured product data as the foundation for all AI discovery surfaces. This means complete and consistent product attributes (size, color, material, compatibility), accurate real-time pricing and inventory, rich product descriptions that answer common consumer questions, and schema markup that helps AI systems parse product pages. Investing in data quality pays dividends across every AI platform simultaneously, while platform-specific optimizations have diminishing returns as the landscape fragments further.

Google took nearly 20 years to teach retailers how to write product titles and descriptions that rank. The AI shopping era is compressing that learning curve into months, and most retailers haven't started. The retailers who will win across all five surfaces are the ones treating product data as infrastructure, not as a marketing task that gets updated quarterly. Your catalog needs to be as real-time and reliable as your payments stack. Every attribute matters because you don't control which attribute an AI will use to make its recommendation.

Meta entering the race confirms what this week made abundantly clear. The AI shopping landscape is fragmenting, not consolidating. Betting on one platform is a losing move. Building a product data foundation that works everywhere is the only strategy that scales. That's exactly what Agentic Commerce Optimization addresses - making your catalog AI-ready across every surface simultaneously.