When a shopper types “a cute travel bag for a weekend trip” into an AI search tool, they do not get ten blue links to scan. They get a short list of products, with prices, reviews and availability, assembled before they ever reach a store. For online retailers, that moves part of the competition away from your page copy and toward the data behind it. Sarkar SEO sees this as the most practical corner of AI SEO, because so much of it is fixable with discipline instead of guesswork.
Shopping is getting AI answers quickly
The change is recent and fast. One analysis of more than 20 million shopping results found that Google AI Overviews rose from 2.1% of shopping queries in November 2025 to 14.0% in March 2026. That does not mean 14% of every ecommerce search. As one summary stresses, the figure does not mean AI Overviews appear on 14% of every ecommerce search on Google. Still, a sevenfold rise in four months is a signal worth acting on. meanecommercefastlane
What Google tells merchants to do
Google’s May 2026 guide for generative AI features has its own section for stores. A summary of it reports that Google recommends Merchant Center feeds and Google Business Profiles for product and local business visibility in AI responses.
The guide’s position on structured data deserves careful reading. It says structured data is not required for generative AI search and that no special schema.org markup needs adding. That is not an instruction to skip product markup. Product markup still feeds regular shopping features, and Google recommends providing both Product structured data and Merchant Center data when possible. Think of it as good hygiene for your store, not a trick for AI. searchenginejournalecommercefastlane
The feed is as important as the page
For shopping assistants, the product feed is often the first thing read. One ecommerce guide argues that product feeds are becoming as important to AI commerce as web pages, because assistants increasingly need structured price, inventory and availability data.
Treat consistency as the main goal:
- Prices, stock status and variants should match between the feed, the page and the checkout.
- Product titles should state what the item is, with brand, model and key attributes, not marketing slogans.
- Identifiers such as GTINs and brand names should be accurate and complete.
- Images should show the actual product and every variant.
- Shipping and return terms should be visible and match across systems.
Mismatches are costly in a different way from ranking problems. A wrong price or false “in stock” label can get a product disapproved or erode shopper trust.
Other assistants use other routes
Do not assume one fix covers every channel. One source notes that Google Search, Shopify Catalog, ChatGPT, Microsoft Copilot and other AI shopping channels use different ranking systems and data sources.
ChatGPT shows how tangled this is. A secondary report says ChatGPT searches across multiple sources, including Google Shopping feeds, Bing’s web index, product review sites, Reddit discussions and merchant websites. The same report cites a March 2026 study of 43,000 products finding that 83% of ChatGPT product recommendations came from Google Shopping’s top 40 organic listings. I could only find that figure through that aggregator, so treat it as a lead to verify, not a settled fact. Even so, one lesson holds: assuming traditional Google rankings guarantee visibility inside ChatGPT or other assistants is a mistake listed among common errors in one ecommerce guide. 50M Shopping Queries Hit ChatGPT Daily and Most Brands Don’t Show Up +2
Platform integrations matter too. Shopify product data is now integrated into ChatGPT through Shopify Catalog, while shoppers complete purchases through the merchant’s own Shopify checkout. If you sell on a platform with such a feed, check your settings and your catalog quality.
Write product pages an assistant can quote
Models favor pages that answer buying questions plainly. For each priority product, add what a careful shopper would ask:
- Who it suits and who should skip it. “Best for carry-on travel, too small for a laptop over 15 inches.”
- Specifications in plain text. Dimensions, materials, compatibility, capacity, warranty.
- Real evidence. Original photos, measured results, or hands-on notes beat manufacturer copy.
- Comparisons. How this compares with your own alternatives, with honest trade-offs.
- Practical terms. Delivery times, returns, sizing help, care instructions.
- Reviews with substance. Detailed reviews that mention use cases help more than star ratings alone.
Manufacturer descriptions pasted across hundreds of stores give a model no reason to prefer yours.
Strengthen the buying guide layer
Category and guide pages matter as the research step. A guide that names selection criteria, explains trade-offs and recommends specific products for specific needs gives AI systems and shoppers something citable. Original testing, even simple, is the difference between a guide and a list. Keep guides current and remove products you no longer sell.
Measure honestly
Measurement is harder in shopping than in plain search. Merchant Center now includes AI helpers such as Merchant Advisor, yet one guide notes that Google has not made AI-specific click data available in Merchant Center. You will need a manual layer:
- Each month, run your top 15 product queries in Google AI Mode, ChatGPT, Perplexity and Gemini.
- Record which brands and retailers appear, which sources are cited, and whether your product is listed with correct price and availability.
- Watch revenue per session and branded search, not only organic sessions.
A starting plan for the next 30 days
Week 1: List your top 50 products by revenue and margin. Audit feed, page and checkout for matching price, stock and identifiers.
Week 2: Fix disapprovals and missing attributes. Rewrite titles that read like slogans.
Week 3: Upgrade the ten most valuable product pages with fit guidance, original images and comparison notes.
Week 4: Run the monthly prompt test and save the results as a baseline. Sarkar SEO suggests keeping the sheet simple enough that anyone on your team can repeat it.
What to avoid
Chasing every channel at once. Start with the feed and pages that serve all of them.
Treating markup as magic. Markup helps systems read your data. It does not compensate for thin content or poor reviews.
Letting data drift. A single sale price missing from the feed can undo weeks of optimization.
