A look at Amazon's and Walmart's differing approaches to agentic AI shopping; Similarweb says Walmart doubled its AI referral share in the US to 32.5% in 2025
Context & Ripple Effects
Walmart’s reported AI-referral position follows a multiyear effort to put AI into the shopping journey, including its generative-AI search rollout and tools for replenishing frequently purchased items. Its later move to combine customer, employee, engineering, and seller agents into four platforms suggests agentic shopping is being built alongside internal and supplier workflows rather than as an isolated chatbot.
The comparison matters because Amazon and Walmart are competing not only for transactions but for where AI-assisted product discovery begins. Referral share is an early measure of that distribution layer, not a measure of total retail sales.
First-order effects
- Similarweb’s 32.5% US referral-share estimate gives Walmart a concrete benchmark in AI-mediated shopping discovery after doubling its share during 2025.
- Amazon faces a clearer competitive comparison: its agentic-shopping approach is being assessed against Walmart’s ability to convert AI-driven discovery into retail traffic.
Second-order effects
- Retailers and brands will have stronger reason to track AI-referral sources separately from conventional search and marketplace traffic, since the source of product discovery may shape which catalogs and offers shoppers see.
- Walmart’s unified-agent architecture can make customer-shopping features more connected to seller and inventory workflows, increasing pressure on rivals to integrate agents across their own commerce operations.
Third-order effects
- If AI referrals become a durable acquisition channel, retail advantage may increasingly depend on control of agent interfaces, product data, and fulfillment access—not just web traffic or marketplace scale.
- The emerging constraint is likely to be permission and trust: commerce agents that can act on shoppers’ behalf will need reliable product information and clear boundaries around recommendations and purchases.
The trend: This is one data point in the shift from search-led e-commerce toward AI agents as a controlled distribution layer for product discovery and transactions.