AI Shopping Statistics 2026: How Consumers Buy With AI

Shopping with AI has moved from novelty to habit in under two years. Consumers now ask ChatGPT, Gemini, Perplexity and retailers' own assistants to research products, compare prices and shortlist options before they ever land on a brand's website.

Salesforce measured a 200% year-on-year jump in shoppers starting their purchase journey with an agentic search, and roughly three to four in ten US and UK consumers now say they use AI somewhere in their buying process, depending on whose definition you accept.

This article covers the most credible 2025 and 2026 data on consumers shopping with AI across all platforms: UK behaviour, generational splits, category strength and retailer adoption. Every figure has been checked against its source.

Key AI shopping statistics for 2026

• 39% of consumers, and over half of Gen Z, already use AI for product discovery, per Salesforce's Connected Shoppers Report.
• Agentic search as the first shopping step grew 200% year on year; AI chat referrals grew 150-428% every quarter from Q1 2024 to Q1 2026, per Salesforce's State of Commerce report.
• 46% of AI users now start purchase research on a standalone AI platform, up from 25% in 2024; traditional search fell from 43% to 24%, per L.E.K. Consulting (via MarTech).
• 39% of US consumers had used generative AI for online shopping by early 2025, with 53% planning to that year, per Adobe's 5,000-consumer survey.
• Only 11% of US consumers will let AI make purchase decisions, even in low-stakes categories, per Gartner.
• UK use of AI assistants for shopping more than doubled in a year, from 12% to 28%, per Adyen's 2026 Retail Report.
• 61% of UK and Ireland consumers have used AI when shopping, yet 68% cannot name one impressive AI shopping experience, per CI&T.
• 77% of UK Gen Z and 75% of Millennials have used AI when shopping, against 29% of Baby Boomers, per Ingrid's Savanta survey.
• 60% of UK consumers would stop using an AI shopping agent after one mistake, per a YouGov survey for ACI Worldwide.
• 86% of US shoppers who used AI for product research verified the recommendation elsewhere before buying, per Product.ai (via MarTech).
• 91% of retail and CPG firms are using or assessing AI, and 89% say AI is increasing revenue, per NVIDIA's State of AI in Retail and CPG survey.
• Agentic shoppers could account for $190bn-$385bn of US ecommerce by 2030, on a NielsenIQ projection (via MarTech).

How many consumers now use AI to shop?

Somewhere between a fifth and half, depending on market and definition. The most-cited figures cluster around 30-40% for people who have used AI at some point in a buying process, with habitual use much lower.

Salesforce found in July 2025 that 39% of consumers were already using AI for product discovery, rising to over half of Gen Z [1]. An Adobe survey of 5,000+ US consumers in March 2025 landed on the same number: 39% had used generative AI for online shopping, and 53% planned to within the year [2].

An L.E.K. Consulting survey of 2,650 US consumers in April 2026, reported by MarTech, put adoption at 30% of all US consumers for informing a purchase decision, whilst Product.ai found 43% of US online shoppers had used an AI assistant for product research in the previous 90 days [3].

The spread between these numbers is definition, not sloppiness. "Ever used AI while shopping" produces figures in the 40s and 50s; "used AI on your most recent purchase over $50" collapses to 20% in Product.ai's data [3].

The practical reading: AI-assisted research is now mainstream behaviour, but it remains one input among several rather than the default shopping channel.

Is AI replacing search as the way shoppers discover products?

It is eroding search's role as the first step rather than replacing it outright. Multiple 2026 datasets show discovery journeys increasingly beginning in an AI chat, whilst traditional search declines as a starting point.

Salesforce's State of Commerce report (July 2026), drawing on behavioural data from over 1.5 billion shoppers in 37 countries, found agentic search as the first shopping step grew 200% year on year. AI chat referrals grew 150-428% in every quarter from Q1 2024 to Q1 2026, whilst overall commerce traffic grew only single to low double digits [4].

The same report found product discovery via traditional search fell 15% between August 2025 and May 2026, brand-owned properties fell 7%, and newer channels, including AI assistants, grew 38% [4].

L.E.K.'s consumer survey tells the same story from the other side: 46% of AI users now start purchase research on a standalone AI platform, up from 25% in 2024, whilst traditional search fell from 43% to 24% as the starting point [3].

Two caveats stop this becoming a "search is dead" headline. L.E.K. measured AI referrals at under 0.5% of total inbound traffic to brand sites, so absolute volumes remain small even at a 176% compound annual growth rate [3]. And Salesforce's finding that 86% of commerce leaders expect large language models to become essential to product discovery within a year [4] is an expectation, not an observation.

What do shoppers actually use AI for when buying?

Research and comparison, overwhelmingly. Shoppers treat AI as a filtering tool that narrows options and surfaces deals, then keep the final decision, and usually the final click, for themselves.

In Adobe's survey, the top AI shopping tasks were research (55% of users), product recommendations (47%), deal-seeking (43%), present ideas (35%), finding unique products (35%) and shopping lists (33%); 92% of AI shoppers said it enhanced their experience and 87% were more likely to use it for larger or complex purchases [2].

L.E.K. found 53% of AI users primarily use it to compare options and narrow a shortlist, ahead of final decision-making (21%) and discovery (19%) [3]. Gartner's January 2026 survey found 31% of US consumers willing to let AI narrow choices for household supplies and 28% for personal electronics, against just 11% willing to let it make the purchase decision [5].

The verification habit is the most under-reported number here. Product.ai found 86% of AI product researchers checked the recommendation elsewhere before buying; Gartner found 54% of recent AI shoppers had to double-check everything the tools told them, and 62% said AI-provided information ended up wasting their time [3][5].

AI is adding a research layer on top of the funnel rather than shortening it, and inaccurate product data in AI answers is now a direct commercial liability.

How many UK consumers shop with AI?

Credible UK estimates in 2026 range from 21% to about 60%, depending on how the question is asked. Habitual, deliberate use sits at the lower end; "have ever used AI whilst shopping" reaches well past half.

Adyen's report shows the growth rate: UK adoption more than doubled from 12% to 28% in a year, with over 18% of Gen Z and 15% of Millennials trying AI shopping for the first time in the previous 12 months [6]. CI&T's broader "ever used" framing found 61%, yet 68% of the same respondents could not name one impressive AI shopping experience [7].

The Commerce and PayPal study captures the intent gap: only 21% of UK shoppers currently use AI shopping tools, but 70% want AI help in future and 62% of non-users plan to try it within a year [8].

The UK picture is early but steepening: casual exposure is widespread, deliberate repeated use is roughly a quarter of shoppers and climbing. UK shoppers also have firm views on commercial influence: 55% want sponsored content in AI tools clearly labelled and 46% say brands should not be able to pay for preference [8].

Which generations shop with AI the most?

Gen Z and Millennials, by a wide margin, in every dataset on both sides of the Atlantic. Boomers are the consistent laggards, and adoption drops sharply after the mid-40s.

Ingrid's UK data shows the starkest spread: 77% of Gen Z and 75% of Millennials have used AI when shopping, versus 29% of Baby Boomers, with high earners on £75,000+ also at 77% [9].

Willingness to hand over the whole journey is less generational than usage, though. Adyen found 49% of UK Millennials would let AI handle the entire purchase, but so would 42% of Gen X and 33% of Boomers [6].

This is not a youth trend brands can wait out. The 27-42 bracket, peak household spending years, is already at three-quarters penetration in the UK on the loosest measure.

Which categories is AI shopping strongest in?

Considered, spec-heavy purchases lead. Travel, electronics and financial products top the research league, whilst electronics and jewellery convert best from AI referrals.

L.E.K.'s tracking found 71% of consumers use AI for travel research, 65% for consumer electronics and 62% for financial products, with no tracked category below 47% [3]. By share of AI-researched purchases over $50, Product.ai found apparel the largest category at 30%, with electronics at 20% [3].

Adobe's transaction data adds the conversion lens: generative AI traffic converted best in electronics and jewellery, and worst in apparel, home goods and grocery [2].

The apparel paradox is worth noting: heavy AI research volume, poor AI-referral conversion, almost certainly because fit and fabric cannot be resolved in a chat. Categories where the deciding attributes are quantifiable should expect AI-referred demand to arrive first.

Do AI-referred shoppers behave differently once they arrive?

Yes. They browse more, bounce less and increasingly arrive with their mind already made up, which is compressing time spent on brand and retailer websites.

Adobe's analysis of over 1 trillion US retail site visits found shoppers arriving from generative AI sources showed 8% higher engagement, browsed 12% more pages and had a 23% lower bounce rate than non-AI traffic. The conversion gap closed sharply too, from 43% less likely to convert in July 2024 to 9% by February 2025 [2].

L.E.K. found 31% of AI users say their purchase decision was largely made before reaching a brand or retailer site, up from 26% two years earlier, and average visit duration fell 6% across 13 categories between March 2024 and March 2026 [3].

The "pre-decided shopper" is the structural change hiding in these numbers. If comparison happens inside the AI conversation, the retailer's site stops being a persuasion venue and becomes a fulfilment venue.

Delivery data is becoming part of that AI-side decision: 56% of UK shoppers in Ingrid's survey would switch products if an AI agent could not show delivery information, and 53% would abandon the session entirely [9]. Abandonment risk is moving upstream of the cart.

Will consumers let AI actually buy things for them?

Mostly not yet. Interest in agentic shopping is high, but willingness to surrender the final purchase decision is low everywhere measured, and tolerance for agent error is close to zero.

Gartner's headline is blunt: willingness to let AI make purchase decisions topped out at 11% of US consumers, even in low-stakes categories [5]. Product.ai found 42% of shoppers would trust AI unverified for a purchase under $50, but only 5% over $500 [3].

UK data mirrors this. Adyen found 44% of UK shoppers open to AI handling the entire shopping process once preferences are set [6], and 64% told Commerce and PayPal they are interested in agentic tools [8]. But the ACI Worldwide YouGov survey of 2,080 UK adults found just 19% trust AI assistants with everyday purchase decisions versus 55% for a human adviser, and 60% would stop using an agent after one mistake [10].

Accountability is unresolved: 54% of UK consumers say the AI company should be responsible for refunds when an agent buys the wrong thing; only 9% blame the retailer [10]. The top barriers in the Commerce and PayPal study were AI purchasing without approval (43%) and bank security fears (39%) [8].

The gap between "interested" (60%+) and "trusting" (11-19%) is the defining tension of agentic commerce in 2026. Vendors are building checkout capability faster than consumers are extending permission to use it.

How quickly are retailers adopting AI shopping features?

Faster than consumers are adopting them, at least on paper. AI use inside retail organisations is near universal, and agentic commerce is the stated priority for 2026 budgets.

NVIDIA's January 2026 State of AI in Retail and CPG survey found 91% of respondents actively using or assessing AI, 90% increasing AI budgets in 2026, 89% reporting AI is helping increase revenue and 95% saying it is cutting costs. On agents, 47% are using or assessing agentic AI, with 20% already live [11].

Salesforce's survey of 3,450 commerce professionals found 28% of organisations using agentic AI and 44% more planning adoption within six months, whilst 75% of retailers in its Connected Shoppers research said AI agents would be essential to compete by 2026 [4][12]. In the UK, Adyen found 84% of retailers open to enabling AI to complete purchases on shoppers' behalf [6].

A readiness gap sits underneath the enthusiasm. Only 27% of organisations in Salesforce's survey have fully unified customer data and only 32% have defined AI success metrics [4]. Fragmented data is how retailers end up feeding AI assistants the inaccurate product information that 54% of shoppers already feel forced to double-check [5].

How to read these numbers

Survey-based adoption figures depend heavily on question wording. "Have you ever used AI whilst shopping" (CI&T's 61%) and "do you currently use AI shopping tools" (Commerce and PayPal's 21%) are both true of the same UK population in the same year.

Measurement types disagree by design. Adobe and Salesforce count observed clicks and transactions across trillions of visits; L.E.K., Gartner and the UK surveys rely on self-reported behaviour from samples of 322 to 5,000 people, where recall bias inflates or deflates figures.

Some respondents do not know when they are using AI. Gartner found 72% of consumers say generative AI appears in their internet use whether they asked for it or not, which muddies any self-reported number [5].

Growth percentages look explosive partly because the base was near zero: L.E.K. puts AI referrals below 0.5% of inbound traffic to brand sites even after a 176% compound annual growth rate [3]. And the NielsenIQ $190-385bn figure, like Salesforce's "essential within a year" finding, is a projection or expectation, not an observation.

What this means going into 2027

Optimise for the pre-decided shopper. With 46% of AI users starting research in a chatbot and 31% arriving at sites already decided, product data accuracy inside AI answers now shapes revenue before your site gets a vote.

Structured product, price, availability and delivery feeds become table stakes for AI surfaces; over half of UK shoppers would abandon a session when an agent cannot show delivery information.

Expect the UK to close the gap on the US quickly. Adoption doubled to 28% in a year on Adyen's measure, and 62% of current non-users say they will try AI shopping within twelve months.

Do not confuse agentic interest with agentic permission. With only 11% of US consumers willing to let AI buy, and 60% of UK consumers ready to drop an agent after one error, checkout autonomy will lag research assistance well into 2027.

Trust is the growth constraint, not capability. The winners will be brands whose AI-surfaced information survives the double-check that 86% of shoppers still perform. If you want a second pair of eyes on how your products currently surface across the AI shopping tools, you can book a call with me.