Are AI Chats Becoming a Starting Point for Product Discovery?
A new Salesforce report suggests that some shopping journeys now begin in AI chats rather than on a retailer’s website or a traditional search page. The findings are useful, but they need careful interpretation.
What the report found
Salesforce published the fourth edition of its State of Commerce research on September 30, 2026. Its platform data indicates that the use of “agentic search” as the first step in a shopping journey grew 200% year over year. Traffic referred from AI chats grew between 150% and 428% year over year in each quarter measured.
Those percentages describe change, not the total share of ecommerce traffic. A fast-growing channel can still begin from a small base. The release also combines global findings with separate Singapore and Asia-Pacific results. Regional figures should not be applied automatically to every market.
Sample, geography and limits
The research includes a double-anonymous survey of 3,450 commerce professionals in 20 countries and 13 industries, conducted from April 10 to June 4, 2026. It also uses a consumer survey of 4,689 shoppers in eight countries from May 12 to 18, 2026.
Salesforce supplemented the surveys with behavioral data from more than 1.5 billion shoppers across 37 countries, covering Q1 2024 through Q1 2026. That scale gives the report a broad view of activity on Salesforce-connected commerce properties. It does not represent every shopper, retailer or AI platform. Because this is observational research, it can show patterns but cannot prove that AI chats caused a change in revenue or loyalty.
Why product data matters
An AI assistant may need to compare size, material, compatibility, stock, delivery and return conditions before suggesting a product. If these facts differ between a product page, feed and marketplace listing, the assistant and the shopper both receive an inconsistent picture.
This is where established ecommerce and SEO work remains useful. Clear product titles, verified attributes, crawlable pages and consistent structured data help systems understand the offer. They also help people make a decision after they arrive. None of these steps guarantees an AI recommendation.
A practical four-step review
- Choose one important category. Start with a manageable group instead of rewriting the full catalogue.
- Compare every source of product facts. Check the page, Merchant Center or marketplace feed, inventory system and customer support answers.
- Answer real buying questions. Add verified details about dimensions, materials, use, delivery and returns. Avoid vague claims that cannot be checked.
- Measure channels separately. Record the change date and compare AI referrals, organic search, direct traffic, add-to-cart actions and sales over a long enough period.
Use the results as a diagnostic signal. A rise in AI referrals may coincide with seasonality, promotions, stock changes or wider brand activity. Compare several indicators before assigning a cause.
Source checked October 2, 2026: Salesforce, “Shopping’s New First Step: Agentic Search Grows 200% as Purchase Journeys Start in AI Chats,” September 30, 2026. The percentages come from Salesforce research and should be interpreted within its stated sample, geography and platform coverage.