According to Adobe Digital Insight, U.S. online spending reached $88.7 billion in October 2025, up 8.2% year over year. Mobile spending was $45.6 billion, or 51.4% of the total, and rose 11.6% year over year. In Japan, e-commerce accounted for about 10% of retail sales, while the share of households of two or more people that shopped online in fiscal 2025 was 56.9%, up 1.6% from the previous fiscal year.

Traffic from generative AI systems to websites grew more than 10-fold between July 2024 and February 2025. Traffic that began with product searches through generative AI and then moved to e-commerce sites rose 1,200% year over year. Adobe reported that these visitors spent longer on sites, viewed more pages and had a 16% higher purchase rate than visitors from other services. In September 2025, generative AI referrals also surpassed other channels in conversion rate, the share of visits that result in an action such as a purchase or inquiry.

Generative AI is beginning to help shoppers find, understand and compare products, beyond serving as a search tool. The article says use is advancing for products that require comparison, expensive goods, investment products, travel, insurance, PCs and home appliances. Another development is shopping agents that interpret buyers’ intentions and select products for them.

Agentic commerce is a system in which AI agents automate complex product selection, business-to-business procurement and reordering everyday goods. U.S. firm AT Kearney defines it as a purchasing and payment model in which AI agents handle product discovery, comparison, negotiation, verification and payment. The firm forecasts that by 2030 the model could account for about one-quarter of annual online sales, moving about $10 trillion to $12 trillion.

Examples include OpenAI’s ChatGPT shopping feature, Perplexity’s Buy with Pro and Amazon’s automatic reordering. ChatGPT takes shoppers’ requirements in conversation and works with partner e-commerce sites to search for and recommend products. Buy with Pro supports the process from search through purchase and can complete payment for some products. The intended next step is for systems to consider purchase history, preferences, budgets, inventory and delivery times, then handle selection, purchase and payment.

As AI takes over gathering shopping information from people, e-commerce sites will need AIO, or information optimization for AI search, so AI search systems cite and recommend their content. Product information must be accurate, reliable, comprehensive, clearly contextualized and promptly updated; inventory and delivery data must also be prepared for use in product selection. Real-time connections for prices, inventory and delivery, as well as structured product data, are needed. Older systems can raise implementation costs, so the article calls for transforming data integration platforms, which can also support demand forecasting and product strategy. The article also says that rising access from generative AI will bring more cyberattacks on APIs, making it essential to remove fake sites and vulnerabilities and defend against DDoS attacks.

According to Adobe Digital Insight, U.S. online spending reached $88.7 billion in October 2025, up 8.2% year over year. Mobile spending was $45.6 billion, or 51.4% of the total, and rose 11.6% year over year. In Japan, e-commerce accounted for about 10% of retail sales, while the share of households of two or more people that shopped online in fiscal 2025 was 56.9%, up 1.6% from the previous fiscal year.

Traffic from generative AI systems to websites grew more than 10-fold between July 2024 and February 2025. Traffic that began with product searches through generative AI and then moved to e-commerce sites rose 1,200% year over year. Adobe reported that these visitors spent longer on sites, viewed more pages and had a 16% higher purchase rate than visitors from other services. In September 2025, generative AI referrals also surpassed other channels in conversion rate, the share of visits that result in an action such as a purchase or inquiry.

Generative AI is beginning to help shoppers find, understand and compare products, beyond serving as a search tool. The article says use is advancing for products that require comparison, expensive goods, investment products, travel, insurance, PCs and home appliances. Another development is shopping agents that interpret buyers’ intentions and select products for them.

Agentic commerce is a system in which AI agents automate complex product selection, business-to-business procurement and reordering everyday goods. U.S. firm AT Kearney defines it as a purchasing and payment model in which AI agents handle product discovery, comparison, negotiation, verification and payment. The firm forecasts that by 2030 the model could account for about one-quarter of annual online sales, moving about $10 trillion to $12 trillion.

Examples include OpenAI’s ChatGPT shopping feature, Perplexity’s Buy with Pro and Amazon’s automatic reordering. ChatGPT takes shoppers’ requirements in conversation and works with partner e-commerce sites to search for and recommend products. Buy with Pro supports the process from search through purchase and can complete payment for some products. The intended next step is for systems to consider purchase history, preferences, budgets, inventory and delivery times, then handle selection, purchase and payment.

As AI takes over gathering shopping information from people, e-commerce sites will need AIO, or information optimization for AI search, so AI search systems cite and recommend their content. Product information must be accurate, reliable, comprehensive, clearly contextualized and promptly updated; inventory and delivery data must also be prepared for use in product selection. Real-time connections for prices, inventory and delivery, as well as structured product data, are needed. Older systems can raise implementation costs, so the article calls for transforming data integration platforms, which can also support demand forecasting and product strategy. The article also says that rising access from generative AI will bring more cyberattacks on APIs, making it essential to remove fake sites and vulnerabilities and defend against DDoS attacks.