Conversational assistants, response engines, and AI-driven purchasing agents are silently transforming the way consumers discover products. A study conducted by Retail Economics, in partnership with AWS, Botify, and DataDome, surveyed 6,000 consumers in France, the UK, and the US, revealing that an increasing portion of commercial decisions is now made before reaching e-commerce sites. The focus for retailers is not just on attracting traffic, but on being visible, understandable, and credible to AI systems that filter the offers. The battle for visibility is now beyond search engine result pages, into algorithmic layers that select, synthesize, and prioritize products before showing a link. For over 20 years, e-commerce operated around a relatively stable model where a consumer expressed intent, conducted a search, compared results, and clicked to a shopping site. E-commerce directions optimized this model through SEO, paid advertising, marketplaces, and affiliation, with traffic as a central indicator. This model is now evolving significantly.
The discovery of products is entering a phase called “agentive.” AI systems now interpret purchase intentions, gather information, compare offers, synthesize recommendations, and, eventually, may complete certain transactions on behalf of the customer. AI no longer just sends links; it filters, organizes, and becomes an active intermediary between the brand and the consumer.
Contrary to expectations, this transformation is not marginal or experimental. According to the study, of the 6,000 consumers surveyed – 2,000 in France, 2,000 in the UK, and 2,000 in the US – 73% stated they had used an AI assistant. Furthermore, 38% had used a third-party assistant like ChatGPT or Copilot for shopping-related tasks, such as getting product ideas, suggestions, or comparisons. 21% used AI to support a purchase decision, and 34% used AI functionalities directly on retailer websites or apps.
Therefore, AI-assisted discovery is now integrated into shopping journeys. Among 18-24-year-olds, about a quarter regularly use AI assistants, with one in five using them daily. In contrast, fewer than one in ten consumers over 55 reported daily usage. The transformation is driven by younger cohorts, giving it a lasting impact.
The most strategic change lies in the shift of influence in the purchase journey. AI platforms extensively analyze product catalogs. By 2025, the traffic from AI bots has increased by 5.4 times. This automated activity surpasses traditional human navigation patterns. Google generates about one visit for every six crawls, whereas OpenAI generates one visit for every 198 crawls.
AI systems collect vast amounts of data but generate relatively little direct traffic. A growing share of product comparison and evaluation now occurs within AI interfaces before users access shopping sites. Value moves upstream of the click: a recommendation from an assistant can shape decisions without full brand control over the context.
In September 2025, after OpenAI introduced new commercial features, visits from this platform to retail sites increased by 200%, accompanied by a significant crawl intensification. Platforms lay the groundwork by managing data before activating transactional functionalities.
Massive crawl automation blurs traditional performance indicators. When a commonly used tracking parameter was removed, some retailers saw a 67% drop in impressions while clicks remained stable. Click-through rates increased by about 150%, suggesting a notable share of impressions came from automated interactions. In an environment where bots can make over a hundred simultaneous requests, digital teams must discern human traffic from machine traffic to avoid basing performance solely on partially artificial signals.
In the new landscape, structured data is pivotal. AI agents prioritize structured, verified, and easily interpretable information. Yet, some e-commerce content, particularly JavaScript-dependent content, remains challenging for automation systems to read fully. Consequently, a product that is visible to humans may remain partially invisible to AI, posing a risk of algorithmic invisibility. SEO remains crucial but evolves toward optimizing for response engines with product attribute structuring, comprehensive taxonomies, detailed metadata, and timely price and availability updates becoming critical factors.
A strategic dilemma emerges: whether to open or control AI access to content. While allowing AI access has visibility, integration, and user experience benefits, there are real risks like malicious scraping, loss of product representation control, infrastructure overload, and fraud. A vulnerability lies in retailers’ ability to identify legitimate AI agents accessing their sites. Malicious programs can mimic legitimate AI agents through “spoofing” the user-agent information to access product pages, retrieve price and stock data, analyze payment pathways, or manipulate traffic indicators.
Sector impacts vary. Retail sectors requiring technical comparisons and rational arbitrage, such as electronics and appliances, are most affected. The intensity of crawling varies: in 2025, bot activity increased by 29 times in food and 11 times in Home & DIY categories, where price and stock volatility attract more automation. Emotional or sensory-driven purchases are likely to evolve more gradually. While AI usage grows, trust is not absolute – 32% of consumers do not trust AI-driven search and discovery. The commerce realm is moving into a hybrid phase, where AI assists but does not entirely replace human decision-making.
Amazon’s AI assistant Rufus exemplifies this potential: by 2025, over 250 million customers had used Rufus, integrated into the retailer’s site and app. Interactions rose by 210% in a year, with Rufus users being 60% more likely to purchase during a session. AI no longer solely explores but becomes a direct conversion lever.
