When a buyer asks ChatGPT "what's the best laptop for graphic design under €1,500?", the answer determines tens of thousands of euros in sales distributed among the cited brands. The e-commerce that doesn't appear in those answers is losing the fastest-growing discovery channel of the moment. This guide explains how to enter.
The new buying journey in 2026
Adobe Digital Economy Index 2026 data shows that 19% of purchase decisions for products >€100 in Spain now go through at least one generative AI query before the final decision. In technical categories (electronics, photography, software), the percentage exceeds 35%. AI acts as a pre-filter: if your product doesn't enter the shortlist the user brings to the e-commerce site, you don't compete.
Schema.org Product: the non-negotiable foundation
The first technical requirement is complete Schema.org Product on each page: name, image, description, brand, sku, gtin13, offers (price, priceCurrency, availability, priceValidUntil), aggregateRating, and review. Spanish e-commerce sites with complete and validated Schema.org Product have a 4.2× higher probability of being cited in AI responses than those without, according to GEOMOND 2026 portfolio analysis.
Price and stock visible in HTML, not only in JS
Many e-commerce sites render prices and availability only in JavaScript after load. AI crawlers (especially PerplexityBot and GPTBot) don't always execute JS. If the price isn't in the initial HTML, the model doesn't see it. Solution: SSR (Server-Side Rendering) or pre-render for product pages, ensuring critical data is in the first HTML served.
Merchant Center feed and its impact on Gemini
Google integrates Merchant Center catalog into Gemini Shopping. Having an updated feed with prices, availability, and quality images becomes a direct signal for Gemini when a user asks about products. Operational action: maintain the feed with minimum daily frequency, especially for prices and stock.
Reviews: the tiebreaker
When there are three products with similar specs, AI chooses by reviews. Three elements count: 1) number of reviews (typical thresholds 50, 200, 1000); 2) average score (minimum 4.2 for consideration); 3) verified reviews on platforms with open API (Trustpilot, Trusted Shops, Yotpo). Reviews on your own domain count less due to lack of independent verification.
Preparation for shopping agents
OpenAI Operator and Perplexity Comet execute purchases on the user's behalf. For an agent to complete the purchase on your e-commerce without abandoning: 1) standard checkout without unnecessary captchas; 2) clear and linked return policy; 3) integration with universal payment methods (Apple Pay, Google Pay, PayPal); 4) load times <3s at each checkout step; 5) Schema.org Offer with shippingDetails and returnPolicy. E-commerce sites failing these points lose sales to competitors after the agent's decision.
Images and multimodal
When a user uploads a product photo and asks "where can I buy this?", multimodal AI identifies the object and suggests e-commerce sites. To enter that suggestion: descriptive alt text on each image, Schema.org ImageObject, and publishing images from multiple angles with neutral backgrounds that facilitate visual recognition.
Real case: fashion e-commerce in Barcelona
A GEOMOND client in women's fashion implemented complete Schema.org Product, daily Merchant Center feed, Trustpilot integration, and SSR for product pages between February and May 2026. Presence Index went from 18 to 47, traffic from generative AI grew 312%, and AI-attributable sales represented 14% of total in month 6, with AOV 22% higher than the SEM channel.
E-commerce is the sector with the highest GEO evolution speed in 2026: each quarter brings new agentic integrations. At GEOMOND we have a specific e-commerce audit module. Request the free audit.
Frequently asked questions
What Schema.org does an ecommerce need to appear on ChatGPT and Gemini?
Product, Offer, AggregateRating, Review and BreadcrumbList are mandatory. Useful additions: ItemAvailability, ShippingDetails, MerchantReturnPolicy. LLMs already read these signals to answer purchase queries.
How do I get an AI agent to recommend my products?
Three requirements: up-to-date structured product feed, verifiable reviews with Schema.org Review, and clear policies (returns, warranty, shipping) accessible without login. Without these, the agent recommends the competitor that has them.
Does GEO replace Google Shopping or is it complementary?
Complementary. Google Shopping captures transactional click intent; GEO on ChatGPT/Perplexity captures pre-purchase comparative intent where your brand must be cited. Both channels reinforce each other.