What: A BCG survey of 50 of the top 300 global retailers and 100 top brands shows most retailers sharing product data with AI agents while protecting customer data, as agents take over discovery but not yet checkout.

Why it is important: Agents are taking over discovery while checkout stays with the shopper, so department stores should invest now in clean product data and in loyalty value agents can read, while keeping the transaction and the customer relationship in their own hands.

Retailers' digital growth has rested on a flywheel in which marketplaces, loyalty and personalization, and retail media reinforce one another. External AI agents such as ChatGPT, Claude and Gemini now sit between retailers and shoppers, so a sale can happen without a visit. In BCG's survey, 81% of US consumers and 53% of European consumers expect to use agents when shopping, yet only 4% would let an agent complete a purchase.

Retailers therefore play two games: keeping direct visits and winning the agent's recommendation. Most share or plan to share product attributes, prices and ratings, but 87% would never share customer data. Only 18% consider their marketplaces ready for agents, and only 14% plan to give agents the same loyalty benefits as their own channels.

BCG warns that agent-routed demand is unlikely to stay free, as online travel agencies now take 15% to 25% of each booking. Its no-regret agenda covers one data policy across the flywheel, agent-readable catalog feeds, loyalty value agents can compare, and retail media that proves incremental sales.

IADS Notes: The caution BCG describes around agent-completed purchases is already visible in earlier coverage. Large online retailers are blocking agents to protect customer relationships and advertising revenue, while consumers use AI for mid-journey research rather than for the final purchase (a16z, October 2026). French data shows the same gap: trust in AI falls from 47% before a purchase to 30% at payment, and Walmart suspended its ChatGPT Instant Checkout integration after conversion came in roughly three times lower than on its own site (Journal du Net, September 2026). McKinsey expects agentic AI to become the default entry point for discovery and routine purchasing, making store visits less frequent but more valuable (McKinsey, May 2026). On the supply side, retailers that cannot structure and govern their product data become invisible to the algorithms that decide what shoppers see first (Journal du Net, March 2026). Marketplaces, which accounted for nearly two-thirds of retail spending by 2024, reward brands that can update products, prices and availability across them quickly (Journal du Net, October 2026).

How retailers can win with AI shopping agents in 2026