AI is becoming a regular part of online shopping, but consumers are not abandoning traditional search. Instead, they are combining the two as they research products and move toward a purchase.
That is one of the main findings of Similarweb’s State of E-commerce 2026 report. Direct referrals from generative AI platforms grew 203% year over year, making AI one of the fastest-growing sources of e-commerce traffic. Yet AI still accounts for only 0.4% of retail e-commerce visits.
The more interesting development may therefore be happening before the click.
Similarweb found that AI recommendations can significantly influence purchasing decisions even when shoppers eventually reach a retailer through another channel. In some cases, a brand recommended by AI can gain as much as a two-to-one purchase advantage over competing brands.
For e-commerce businesses, AI’s influence may be considerably larger than referral traffic alone suggests.
Consumers increasingly use tools such as ChatGPT and Gemini to explore products, understand categories and narrow down their choices.
But AI is not simply replacing Google.
Similarweb reports that 89% of consumers who use AI during shopping research also use search. According to the company, consumers are “stacking” tools rather than switching from one to another.
The combination also appears to produce valuable shopping journeys. Similarweb found that journeys involving both AI and search convert at 23%, 84% higher than journeys involving AI alone.
The two channels can therefore play different roles.
A shopper might ask an AI assistant which type of monitor is suitable for graphic design, compare several recommended models and then use Google to search for prices, retailers or additional reviews before making the final purchase.
In that journey, search may receive the measurable click, while AI helped determine which products entered the consideration set in the first place.
This creates a measurement challenge for e-commerce businesses.
If a consumer first researches a product through an AI assistant but later reaches the retailer through Google, direct traffic, or a marketplace, conventional analytics may attribute little or no value to the AI interaction.
Similarweb’s findings suggest that focusing only on AI referral traffic can therefore give an incomplete picture.
AI may increasingly operate as an influence channel rather than simply another source of website visits.
That distinction matters because product discovery often begins well before consumers decide where to buy. AI can answer questions, explain differences between technologies, compare specifications, and recommend particular brands or products without sending the shopper directly to a retailer.
For brands, being included in those conversations could become an important part of digital visibility.
The change also raises a practical question: what information does an AI system have available when a shopper asks it to compare products?
This is particularly relevant to Icecat’s ecosystem.
AI-assisted product research depends on detailed information about the products being considered. Specifications, identifiers, dimensions, compatibility, features, descriptions, and product relationships can all help distinguish one option from another.
Structured product information gives brands, retailers and marketplaces a consistent way to describe those products across channels.
That becomes particularly valuable when product information is no longer consumed only on a product detail page. The same information may need to support marketplace listings, search engines, conversational assistants, and eventually AI agents that can take action for consumers.
If a product’s specifications are incomplete, inconsistent, or hard for machines to interpret, AI systems may struggle to understand where that product fits within a shopper’s requirements.
Product content therefore increasingly supports both human and machine-led discovery.
AI is not the only change identified in Similarweb’s report.
Overall ecommerce website traffic increased 6.8% year over year, while ecommerce app sessions are growing at roughly 1.3 times the rate of web visits. In the United States, 86.5% of consumers say they primarily shop using a smartphone or tablet.
The shopper journey is therefore becoming more fragmented across interfaces rather than moving toward one dominant channel.
A consumer might discover a product through social media, ask an AI assistant about it, search for alternatives on Google, open a marketplace app, and finally buy it on a smartphone.
For e-commerce businesses, maintaining consistent product information across these environments becomes increasingly important. The shopper may change channels several times, but the underlying product should remain recognizable and accurately described throughout the journey.
For years, e-commerce optimization concentrated heavily on bringing consumers to a website or marketplace product page.
Similarweb’s findings show why businesses may need to look further upstream.
One in four U.S. shoppers now uses AI before buying, the report finds. Nearly half say AI has already changed how they browse for products.
AI may still generate relatively little direct e-commerce traffic, but it is becoming part of the decision-making process that determines what consumers eventually search for and buy.
That changes the role of product content.
Accurate, structured product information isn’t only about creating a better product page after the shopper arrives. It can also help make products understandable across the growing number of systems involved before the visit.
Search remains important, apps continue to grow, and AI is adding another layer to the journey rather than replacing what came before.
For brands and retailers, the challenge is no longer simply being present at the final point of purchase. Their products must remain understandable and discoverable throughout an increasingly complex path to purchase.
Read further: News, AI, e-commerce, ecommerce, Icecat, product content
