SourceNode research

Agentic commerce in 2026: what is live, what the data shows, and how products get chosen

Agentic commerce in 2026: what is live, what the data shows, and how products get chosen

Agentic commerce in 2026: what is live, what the data shows, and how products get chosen

53% of consumers used AI for shopping in the past three months, but only 16% would let it complete a purchase. What is actually live, and what decides which products an assistant shows.

53% of consumers used AI for shopping in the past three months, but only 16% would let it complete a purchase. What is actually live, and what decides which products an assistant shows.

Odin Lund

Research, SourceNode

US consumers and AI shopping, 2026
Used AI for shopping in the past three months48%
Use generative AI as a primary holiday shopping tool39%
Would trust AI to complete a purchase16%
Source: McKinsey, June 2026. Deloitte, September 2026
US consumers and AI shopping, 2026. Source: McKinsey, June 2026. Deloitte, September 2026
Used AI for shopping in the past three months48%
Use generative AI as a primary holiday shopping tool39%
Would trust AI to complete a purchase16%

Key findings

  • 53% of consumers across 11 markets used AI for shopping in the past three months. McKinsey, 2026.

  • Only 16% of US consumers would trust AI to complete a purchase with their permission. Deloitte, 2026.

  • OpenAI moved away from checkout inside ChatGPT in March 2026. Purchases now finish on the merchant’s own site. OpenAI.

What is agentic commerce, and how big is it?

Agentic commerce is shopping where an AI assistant does part of the work for a person: researching, comparing, building a basket and, in some cases, paying. The assistant can be a general one such as ChatGPT or Gemini, or a retailer’s own.

The forecasts are far apart, which tells you how early this is.

  • McKinsey, October 2025: up to $1 trillion in US retail revenue orchestrated through agents by 2030, and $3 trillion to $5 trillion globally. McKinsey’s term is orchestrated revenue, which is wider than sales completed by an agent.

  • Morgan Stanley, December 2025: $190 billion to $385 billion of US e-commerce spending by 2030, or 10% to 20% of the market.

  • eMarketer, July 2026: $19.74 billion in US sales driven by AI platforms in 2026, a forecast it had just cut by about 4%.

The near-term number is small and the long-term range is wide. Plan for the direction, not the figure.

What is actually live in November 2026?

  • ChatGPT. OpenAI launched Instant Checkout in September 2025 with the Agentic Commerce Protocol, built with Stripe. In March 2026 it moved away from standalone in-chat checkout. Shoppers now discover and compare products in ChatGPT and complete the purchase on the merchant’s own site. The protocol now carries product feeds, and Shopify and Etsy catalogs are already integrated.

  • Google. Agentic checkout began rolling out in November 2025. The Universal Commerce Protocol followed in January 2026, developed with Shopify, Etsy, Wayfair, Target and Walmart. On 15 October 2026 Google began rolling out Universal Cart in the US across Search and the Gemini app, with checkout running on that protocol.

  • Microsoft. Copilot Checkout launched in the US in January 2026 with PayPal, Shopify and Stripe.

  • Perplexity. Instant Buy launched with PayPal in November 2025.

  • Amazon. More than 300 million customers used its Rufus assistant in 2025, and in May 2026 it became part of Alexa for Shopping. Amazon also went to court to keep Perplexity’s agent out of logged-in parts of its site. An injunction granted in March 2026 was lifted on appeal in August.

  • Payments. Visa Intelligent Commerce and Mastercard Agent Pay were announced in April 2025, and Google’s Agent Payments Protocol in September 2025.

This changes monthly. The ChatGPT reversal is the clearest example, so check the date on anything you read about checkout inside an assistant.

Will people let AI buy for them?

They use it to decide far more than to pay.

  • Research is mainstream. 53% of consumers across 11 markets used AI for shopping in the past three months, and 48% in the US. McKinsey, about 21,000 consumers, June 2026.

  • Buying is growing from a low base. Among US AI users, those who completed a purchase with AI rose from 10% to 17%. Same McKinsey study.

  • Trust to pay is limited. In Deloitte’s 2026 holiday survey of 4,000 US consumers, 39% used generative AI as a primary shopping tool and 16% would trust it to complete a purchase with their permission. Gartner found 11% willing to let AI make the purchase decision, from 322 US consumers. Forrester reported that three quarters of consumers were uncomfortable letting an agent complete a purchase.

  • The traffic is real. Visits to US retail sites from generative AI tools rose 693.4% in the 2025 holiday season, and in September 2026 they converted 90% better than other traffic. Adobe expects another 130% rise this holiday season.

So the decision is already being made inside the assistant, even when the payment happens on the store. Being in the answer matters before checkout does.

What decides which products an assistant shows? And sources

The platforms say much the same thing in their own documentation.

  • Results are not ads. OpenAI says ChatGPT selects product results independently, using structured data from first-party and third-party providers and other third-party content. Merchants are ranked on availability, price, quality and whether they make or are the primary seller of the item. Review summaries come from public websites.

  • A feed gives control, not a guarantee. OpenAI’s feed specification says eligibility does not guarantee display. Shopify says the same of its catalog, and that each AI channel controls its own ranking.

  • Checkout has entry rules. Google’s protocol requires a Merchant Center account in good standing, a return policy with cost and window, customer support details and an opt-in on each product.

  • Reviews count. In McKinsey’s 2026 survey, review websites were the second biggest influence on AI shoppers’ most recent purchase, at 12%, behind AI tools themselves at 29%.

The practical list is short: complete and current product data, identical facts on your site and at every retailer, clear policies, and third-party reviews the assistant can read. How SourceNode does this for retail and e-commerce.

The researcher

Odin Lund is co-founder and COO of SourceNode and leads its data analytics and research, including the daily tracking behind this work across ChatGPT, Gemini and Google AI Overviews. Odin founded First Query, the AI search practice that became SourceNode, and co-founded The Pixel Paper.

Meet the research team

About this research

Published by SourceNode

SourceNode is a data and research-first AI search firm. We record what AI models say about a business every day, trace each claim to its source, and use that data to get businesses named in AI answers. Every figure we publish carries its method and its sample.

Share Blog