
AI has doubled its place in the customer journey
in two years
Similarweb and Statista have published the report The State of Ecommerce 2026. It combines behavioral data from millions of shopping journeys with large consumer surveys. Here are the key findings on how AI is used when people shop online.
The report is one of the most comprehensive attempts so far to measure what AI actually does in the customer journey, beyond what people say they do. It builds on four data sources: Similarweb’s traffic and clickstream data, data on products and assortments from online stores and marketplaces, a dedicated consumer survey and Statista’s consumer and market data. The main figures apply to the US, with comparisons against the UK and Germany where data exists.
In June 2026, generative AI was part of 11.4 percent of all shopping journeys that ended in an online store. In July 2024, the share was 4.5 percent.That is more than a doubling in under two years.
The curve has nevertheless leveled off. The share was already 10.9 percent in October 2025 and has stayed around 11 percent since.
Consumers notice the change themselves. Almost half of Americans (49 percent) say that AI has changed the way they use the internet.Among those aged 25 to 44, it is over half. In every age group, the share is above 40 percent.
The clearest finding in the report is that AI is rarely used alone. In 89 percent of shopping journeys where AI is involved, the customer also uses a search engine. Only 11 percent use AI without search.
By Marius Vestlien Hansen
SEO/SEM Lead, Maverix
marius@maverix.no
The journeys where the customer uses both tools are the most extensive. The customer then visits 14.9 websites on average before purchase, against 10.1 for those who use only search and 7.3 for those who use only AI.
These are also the journeys that most often end in a purchase. 23 percent of the journeys with both AI and search convert. That is 5.7 percentage points higher than for search alone and 10.5 percentage points higher than for AI alone. Similarweb stresses that the figure shows a correlation. It may indicate that these are the most carefully considered purchases, rather than that more touchpoints lead to more sales.
Similarweb has divided each shopping journey into the first touchpoint, intermediate touchpoints and the last touchpoint before the online store. The pattern is clear:
• AI is an intermediate touchpoint in 76 percent of the journeys where it is involved. It is the first touchpoint in 23 percent and the last in 18 percent.
• Search is evenly distributed: first touchpoint in 50 percent, intermediate in 60 percent and last in 56 percent.
AI does most of the work in the research and evaluation phase, while search is present all the way.
The role is changing, though. Since July 2024, the share where AI is an intermediate touchpoint has fallen from 81 to 76 percent. At the same time, AI has more often become the first touchpoint (from 19.5 to 23.2 percent) and the last touchpoint (from 14.5 to 18.4 percent). More people start the journey with AI, and more return to AI right before they buy.
The consumer survey shows the same. When people want to find ideas or compare alternatives, nearly twice as many think AI is most useful as think search is (34.5 against 18.9 percent and 36.0 against 19.3 percent respectively). When they want to assess value and feel confident in the choice, the two are almost equal (29.2 against 27.3 percent).
The report also measured what happens after the customer has received a recommendation from AI. Within seven days, users were clearly more likely to visit the brand AI recommended than a competitor that was not recommended.
FFinance: Capital One 14.2 against 3.8 percent, American Express 7.2 against 3.1 percent
Travel: Kayak 12.0 against 3.4 percent, Skyscanner 9.5 against 7.6 percent
Beauty: Sephora 7.9 against 3.3 percent, Ulta 7.6 against 4.6 percent.
Similarweb writes that AI is unlikely to decide a purchase on its own, but that it can help determine which brands the customer carries forward.
How much AI is used varies with what people shop for. The share of shopping journeys with AI is highest in consumer electronics (14.1 percent) and beauty (13.2 percent). Fashion follows (11.7 percent), then marketplaces (11.0 percent) and home and garden (10.4 percent). The report explains the difference by saying that electronics and beauty require more research before purchase.
The report also shows what people ask ChatGPT about, and which websites get the most traffic from AI compared with regular traffic:
Electronics: Questions about PCs, laptops and accessories dominate, often about compatibility and specifications. The online store Newegg is number 4 in AI traffic, but number 9 overall. Ring and Roku rank three places lower in AI traffic than overall.
Beauty: Hair care and skin care dominate, and AI is used to build routines. La Roche-Posay is number 10 in AI traffic against number 75 overall, and The Ordinary is number 8 against number 36.
Fashion: AI is used as a personal stylist. Abercrombie climbs 29 places and Quince 4, while Shein falls 4 places.
Home: AI is used as an interior designer. Pottery Barn and Crate & Barrel climb 5 places each, while the home improvement chain Menards falls 4.
At the top of the lists, it is often the same ones that get the most traffic overall, for example Apple, Best Buy and Samsung in electronics, Ulta and Sephora in beauty and Home Depot and Lowe’s in home and garden.
Traffic that comes from AI tools still makes up under one percent of visits to online stores, both overall and in each individual industry. For e-commerce overall, the share is 0.4 percent. The channel is growing by 203 percent a year, though, and by between 200 and 280 percent across the industries. That is six to ten times faster than the second-fastest channel.
By comparison, 44 percent of visits come directly and 23.9 percent from organic search. Organic search grew by 1.0 percent over the past year. Paid search grew by 22.8 percent and display by 26.8 percent. In fashion, paid channels now account for 33.8 percent of visits and have overtaken direct traffic.
Similarweb’s point is that the growth rate, not the share, is the number to watch.
Part of AI use never happens in ChatGPT or Gemini, but inside the marketplaces themselves. Traffic from Amazon’s own AI assistant, Alexa for Shopping (formerly Rufus), to product pages grew by 537 percent in one year. That is faster than the growth in traffic from external AI tools.
The report points out that these assistants have their own rules for who gets shown, and that they put heavy weight on sponsored placements and the customer’s past purchases. Customers also ask for combinations of properties and uses rather than single words. That places greater demands on the product data.
Statista’s surveys in the US, the UK and Germany show that 21 to 26 percent use AI to research products online.. 43 percent of Americans used ChatGPT in the past month. 27 percent of Americans often use AI as part of the shopping journey, while 44 percent of Germans say they never do.
Trust is moderate. Around one in three is comfortable letting AI narrow down the options, and around four in ten trust AI to recommend products that fit. Only 21 percent would let AI make the final decision. Trust is highest in the US and lowest in Germany.
Younger consumers, men, people with children in the household and people with high incomes use AI most often when they shop. The willingness to let an AI agent buy on their behalf is greatest for clothing and electronics and lowest for luxury goods. In the UK, 39 percent of Generation Z say they are comfortable letting AI book travel for them.
The report describes how AI tools are moving from helping the customer find products to completing the purchase themselves. Google’s Universal Commerce Protocol and OpenAI’s Agentic Commerce Protocol make it possible to find, evaluate and buy a product without leaving the AI tool. That makes it harder to measure what AI actually means for sales. Similarweb and NIQ are therefore planning a new measurement tool for this in the fourth quarter of 2026.
Statista estimates that AI in the customer journey accounted for almost two percent of the growth in global e-commerce revenue from 2024 to 2025. Global B2C e-commerce revenue is expected to pass $4.9 trillion in 2030, 27 percent more than in 2026.
Similarweb itself sums it up: AI sits on top of the customer journey and leaves it in place. Customers keep their tools and use several at once. The report therefore argues that visibility in AI and visibility in search should be treated as one field, and it points to three things that decide the outcome: product data detailed enough for an AI tool to reason over, images and descriptions made for both humans and language models, and a consistent brand across channels.
The figures apply to the US, the UK and Germany. In Norway, the use of generative AI is higher than in any EU country: 56 percent of Norwegians between 16 and 74 used such tools in 2025, against 32.7 percent on average in the EU. There is little to suggest that the patterns in the report are less relevant here.
Read more in the glossary about AI visibility, AI search and GEO.

