Benchmarks

AI agents compare products by weighing price, specifications, reviews, and availability data with a hierarchical scoring system that prioritizes objective attributes over marketing claims. In our analysis of 50,000 product comparison queries across ChatGPT, Perplexity, and Google AI Mode, stores with structured comparison schema see 47% higher citation rates than those relying on unstructured product descriptions. How AI Agents Compare Products: The Evaluation Framework Modern AI shopping agents don’t simply scrape product pages and list them. They build structured comparison matrices by extracting and normalizing product attributes from multiple sources. The process works in three phases: attribute extraction, cross-product normalization, and weighted scoring. ...

Three-quarters of enterprise leaders say they are adopting agentic AI. Only a small fraction have it running in meaningful production. That gap between ambition and reality is the defining feature of ecommerce in mid-2026, and it determines which stores capture AI-driven sales and which get locked out. The infrastructure for agentic commerce is arriving faster than most stores can absorb it. Google’s Universal Cart, rolling out across Search and Gemini in summer 2026, lets shoppers add products from any merchant into a single intelligent cart and checkout via Google Pay. The Universal Commerce Protocol (UCP) is expanding to Canada, Australia, and the UK. The Agent Payments Protocol (AP2) gives AI agents the ability to complete purchases on a customer’s behalf with strict guardrails. But on the merchant side, most stores cannot be found, compared, or purchased by these agents because their product data, schema, and checkout infrastructure are not ready. ...