Feed Management for AI Shopping: Beyond Google

The Feed Landscape in 2026

For years, merchants managed one product feed for one destination: Google Shopping. You uploaded a spreadsheet to Google Merchant Center, optimized your titles and images, and that was the extent of feed management. That era is over.

Three AI shopping surfaces now matter, and each consumes product data in a fundamentally different way. Understanding these differences is not optional if you want your products to appear where shoppers are increasingly going to buy.

Google AI Mode

Google AI Mode pulls from your existing Merchant Center feed combined with on-page structured data from your website. It synthesizes product comparisons directly within a conversational interface, showing users attribute-level comparisons rather than simple product tiles. There are currently no transaction fees for purchases attributed to AI Mode, making this a purely organic channel. The data pipeline is familiar to any merchant already using Google Shopping, but the way that data gets presented to users is radically different.

ChatGPT Shopping

ChatGPT Shopping has its own OpenAI Product Feed specification, separate from Google. It accepts TSV, CSV, XML, or JSON formats. The feed is the primary trusted data source for ChatGPT product recommendations, not crawled web data. This feed powers Instant Checkout, which carries a 4% transaction fee processed through Stripe's Agentic Commerce Protocol. A critical distinction: ChatGPT Shopping results are not ad-influenced. There is no paid placement. Ranking is entirely organic, driven by feed quality and product relevance.

Perplexity

Perplexity takes a hybrid approach. It crawls product pages and also accepts submitted feeds through distribution partners. Its model centers on source attribution, meaning users see exactly which merchant provided the product data alongside each recommendation. Checkout is handled through PayPal integration, and Perplexity's Comet browser (launched July 2025) includes a built-in AI shopping assistant that makes the platform a direct competitor for shopping queries.

The challenge is clear: each platform has different requirements, different submission methods, and different signals that boost product visibility. Most merchants are feeding zero of these platforms intentionally. They have a Merchant Center feed because Google Shopping required it, but they have not considered how that same data could reach ChatGPT or Perplexity. For a broader view of how these platforms fit together, see our complete guide to AI shopping platforms.

The OpenAI Product Feed Spec

The OpenAI Product Feed is the newest and least understood feed specification in e-commerce. Unlike Google Merchant Center, which has been refined over a decade, the OpenAI feed is relatively new and many merchants have never seen its documentation. Here is what you need to know.

Required Fields

Every product submitted to ChatGPT Shopping must include these fields: product_id, title, description, link, image_link, price, availability, brand, and gtin. Missing any of these results in the product being rejected from the feed entirely. The GTIN requirement is worth highlighting because many merchants treat GTINs as optional. For ChatGPT, they are mandatory. If you do not have GTINs for your products, see our GTIN product identifiers guide for how to obtain them.

Recommended Fields

Beyond the required fields, OpenAI recommends including additional_image_links, sale_price, shipping_weight, product_type, and google_product_category. These fields are not required for feed approval, but products with them populated consistently outperform products without them in ChatGPT's ranking algorithm.

Performance Signals That Boost Ranking

This is where ChatGPT's feed diverges most from Google's. OpenAI's specification includes performance signal fields that directly influence which products ChatGPT recommends: popularity_score, return_rate, review_count, and average_rating. These signals are unique to ChatGPT Shopping. A product with a 4.6 average rating across 2,000 reviews and a low return rate will rank significantly higher than a product with no review data, all else being equal. These fields give ChatGPT a trust signal that the product delivers on its promise.

Rich Media Support

The OpenAI feed also accepts video URLs and 3D model links. Products with video content receive higher visibility in conversational shopping flows because ChatGPT can present richer product experiences to users. If you already produce product videos for your website or social channels, including those URLs in your ChatGPT feed is a low-effort, high-impact optimization.

No Pay-to-Play

The most important strategic difference between the ChatGPT feed and Google Shopping: there is no advertising layer. ChatGPT Shopping is not ad-driven. You cannot pay for higher placement. Ranking is purely based on feed quality, product relevance to the user's query, and the performance signals described above. This makes ChatGPT Shopping a genuinely organic channel where small merchants compete on equal footing with large retailers.

Submission and Refresh

Feeds are submitted through OpenAI's commerce portal. The key advantage is refresh frequency: your feed can be updated as often as every 15 minutes for real-time inventory and pricing sync. This is significantly faster than Google Merchant Center's typical processing cadence. Merchants selling products with volatile inventory or frequent price changes benefit enormously from this rapid refresh cycle.

For detailed platform-specific instructions, see our guides on ChatGPT Shopping for merchants and getting products into ChatGPT Shopping.

Google AI Mode vs. Traditional Google Shopping

Google AI Mode and traditional Google Shopping share the same Merchant Center data, but they present it to users in completely different ways. Understanding this distinction changes how you think about feed optimization.

Traditional Google Shopping

The format merchants know well: product tiles arranged in a grid, each showing an image, title, price, and merchant name. Success in traditional Shopping depends heavily on your product image quality, title optimization for click-through, and whether you are running paid Shopping ads. The format is visual and competitive, with products fighting for attention in a crowded grid.

Google AI Mode

AI Mode takes the same Merchant Center data and synthesizes it into a conversational response. Instead of showing a grid of products, AI Mode might tell a shopper: "Based on your requirements, here are three options. Product A from Brand X is the lightest at 280 grams with a 2-year warranty. Product B from Brand Y is 30% heavier but includes free expedited shipping. Product C from Brand Z has the highest customer rating at 4.7 stars across 3,200 reviews."

What This Means for Your Feed

Comparative attributes matter dramatically more in AI Mode. Materials, dimensions, warranty terms, weight, compatibility specifications, and detailed product features are now directly compared in conversations. The attributes you used to skip because Google Shopping never displayed them on product tiles are now getting surfaced word-for-word when the AI compares your product against competitors.

An AI agent might tell a user that your product includes a 2-year warranty while a competitor offers only 90 days. Or that your product is made from recycled materials while alternatives are not. Or that your product weighs 30% less than the next closest option. These comparisons only happen if the data exists in your feed.