Building websites that AI can shop

More than half of consumers (58 percent) have replaced traditional search engines with generative artificial intelligence tools for product recommendations. That figure was 25 percent just two years earlier. Your next storefront visitor may not be a person at all. It may be a large language model (LLM) deciding whether to recommend your products.
Most product pages were never built for this kind of visitor. They were designed around human browsing patterns, optimized for Google’s crawlers, and organized by categories that make sense to shoppers clicking through menus. AI engines operate differently. They parse, embed and synthesize. They look for semantic clarity, structured relationships between entities, and factual density. A product page that works for a human shopper may be completely invisible to ChatGPT, Perplexity, or Google’s AI Overviews.