AI is becoming a familiar shopping companion. Amid the broader wave of AI adoption (both ChatGPT and Gemini crossed more than a billion monthly active users 1 2), consumers are clearly turning to AI to search for products, compare choices, and decide what to buy, with tools like Muse and Alexa for Shopping gaining traction.3
Even as broader AI-driven shopping grows, early signs of a second transition from AI-assisted decisions to AI-executed transactions are emerging. That finding, among others, comes through loud and clear in the recent McKinsey Agentic Commerce Consumer Survey on AI-enabled consumer shopping behavior.
In earlier articles, McKinsey has shone a light on what agentic commerce is, and how it can adapt. This analysis focuses on the 61 percent of global consumers surveyed who are AI users (those who have used consumer AI tools in the past three months) to help illuminate how agentic commerce is evolving and what companies can expect.
Our analysis is based on consumer responses to a June 2026 survey (see sidebar “Research methodology”). Behaviors are continuing to shift with the pace of AI adoption and this analysis is just one source among many that leaders should consider when evaluating consumer behavior. However, as an indicative trend, the analysis can be helpful and highlights that AI users are increasingly relying on AI across the shopping experience, from search through consideration to purchase. And it identifies the emerging gap between consumer readiness and commerce infrastructure.
Of the many findings in this analysis, business leaders should pay special attention to these five:
- More than 50 percent of global consumers now use AI for shopping.
- In the United States, the rate at which AI users completed a purchase with AI jumped 70 percent over last year’s rate.
- Retailer AI assistants trail stand-alone AI tools by as much as two times throughout the shopping journey but catch up at checkout.
- For AI users, AI search is three times more influential than traditional search.
- Agentic commerce is helping e-commerce laggards make up a lot of ground.
Our findings suggest that agentic commerce is reshaping the online shopping experience. As chief marketing officers (CMOs) in particular consider the implications of these trends, one point deserves emphasis: Agentic commerce is not just another e-commerce channel; it is the new engine of e-commerce (see sidebar “Agentic commerce terms”).
A closer look at the analysis
McKinsey analysis has highlighted five important trends.
More than 50 percent of global consumers now use AI for shopping
Across the countries we surveyed, 53 percent of consumers on average have used AI for shopping in the past three months (Exhibit 1). China and India together are about 29 percentage points ahead of the average of other countries when it comes to adopting AI for shopping, driven by high use of AI tools and assistants. India leads all surveyed countries in use of the global AI tools, with 77 percent of its AI users saying they use Amazon’s Alexa for Shopping versus 46 percent of the average (excluding China). China boasts its own rich landscape of domestic AI shopping offerings (for example, Alibaba’s Qwen and DeepSeek).
On average, about half of consumers now shop with AI, although consumers in China have the highest adoption rates.
1Share of adults aged 18+ who used AI chat/search tools or retailer AI assistants for at least 1 shopping task in the past 3 months; representative sample of 11 markets, weighted, n = 9,988; includes 110 respondents aged 70+; markets pooled in proportion to sample size.
Despite leading in private AI investment, US consumers are behind the global average, with just 48 percent having used AI for shopping. Some 74 percent of US AI users trust AI chat and search tools, which is higher than the global average but well below the leader, India, at 87 percent.
Gen Z and millennials are leading globally in the rates for AI search adoption for shopping (62 percent of Gen Z and 63 percent of millennials used AI for shopping versus the 53 percent average for all users). Gen Z in the United States, however, trails its global peers, with 57 percent adopting AI for shopping. Despite Gen Z’s relatively high usage of AI, Gen Zers tend to be more skeptical of AI on average (67 percent of Gen Z AI users trust AI chat and search tools versus the global average of 73 percent), though that varies by country.
Countries with the highest AI adoption rates see that trend carry over across generations. Gen X and baby boomers in China are adopting AI for shopping more than Gen Z and millennials in other markets (Exhibit 2).
Across surveyed markets, Gen Zers and millennials are more likely to use AI for shopping than older consumers.
1Share of adults aged 18+ who used AI chat/search tools or retailer AI assistants for at least 1 shopping task in the past 3 months, by generation; representative sample, weighted, n = 5,822 across the 7 markets shown, including 65 aged 70+.
2Fewer than 100 respondents; indicative.
In the United States, the rate at which AI users completed a purchase with AI jumped 70 percent over last year’s rate
In the United States, AI users who made a purchase with AI grew from 10 to 17 percent year over year on average across all categories we measured in December 2025 (corresponding to 5 to 10 percent of the general US population). Globally, 14 percent of AI users reported using an AI tool to complete a purchase (corresponding to 9 percent of the general population). While that represents a small percentage share of retail gross merchandise value, the growth is notable. Across categories, AI was used to complete purchases for 7 to 18 percent of AI users, with apparel purchases leading the way (Exhibit 3).
Consumers are increasingly turning to AI to complete purchases, most often in apparel or grocery.
by category, past 3 months,1 % of respondents
1Question: How have you used AI tools or assistants (eg, ChatGPT, Gemini, Perplexity) when making purchases in the following categories? Completed a purchase directly within an AI tool once, or more than once or regularly. All AI users across 11 markets, weighted, n = 11,088. US change vs Dec 2025: US respondents only.
2N/A; category not asked in December.
AI users seem eager to let AI do more of the buying. In the top six purchase categories, between 21 and 30 percent of AI users said they would let AI complete a purchase, roughly 1.5 times as many as those who have made an actual purchase with AI. Retailers see the demand and are moving quickly to implement AI shopping capabilities, through open standards such as the Universal Commerce Protocol stack and by opening their doors to agents in a secure, controlled, and standards-based way as the 2026 holiday season approaches.
The scale of the change needed, however, may constrain the full potential in the short term given the architecture, configuration, and deployment work needed across thousands of global players. Implementation complications, for example, have reportedly led OpenAI to scale back its instant checkout ambitions for the time being.4 Furthermore, retailers are also deciding whether to block bots that are scraping their sites in a less controlled fashion.
Retailer AI assistants trail stand-alone AI tools by as much as two times throughout the shopping journey but catch up at checkout
Customers prefer AI chat and search tools over retailers’ own AI assistants in almost all phases of the shopping journey. Overall, 56 percent of consumers used an AI chat or search tool in the past three months, versus 27 percent who used a retailer’s AI shopping assistants.
When looking more granularly at the shopper journey, two things become clear: One is that consumers generally use AI tools less the more they progress through the phases of the shopping journey (except when they get help on making a decision). Second, the gap between using stand-alone AI tools and retailer AI assistants narrows over that progression. That gap closes completely when an AI-using shopper makes a purchase, where retailer AI assistants are marginally preferred (15 percent) versus AI tools (14 percent) (Exhibit 4).
AI platforms dominate discovery, but their lead disappears by checkout, where retailer assistants are used just as often.
1Share of all AI users, 11 markets, weighted, n = 11,088 (retailer assistant nonusers count as not using). The last point compares 2 differently worded items: AI platforms “completed the purchase for me,” and retailer AI assistants “helped with completing a purchase or checkout.”
Part of the reason that retailer AI assistants outperform stand-alone AI platforms at the point of purchase is that retail AI platforms already have infrastructure in place for live inventory, payments, and delivery.
Among retailers, Amazon’s Alexa for Shopping is the most used AI shopping assistant on average (Exhibit 5). Among stand-alone AI tools, ChatGPT and Gemini are the top two most used in the ten markets outside China, as Gemini has gained share over the course of 2026. In China, ByteDance’s Doubao leads.5
Shoppers use several AI tools, not one, so brands need to be visible across multiple platforms in every market.
Note: Markets are sorted by the sum of their top 3 tools’ shares; China and Mexico are tied at 209. In Japan, Gemini and ChatGPT are tied at 59%. Amazon = Amazon for shopping; Google AI = Google AI Overview/AI Mode.
1Question: “Which of the following AI chat/search tools or shopping assistants built into retailer apps/websites have you personally used in the past 3 months?” AI users, ~1,000 per market (n = 11,087); weighted.
While the main usage for AI assistants is to make a purchase, their value seems to go beyond that. Amazon has said Alexa for Shopping6 users are 60 percent more likely to complete a purchase, and US customers who use Alexa for Shopping spend 40 percent more on average compared with shoppers who don’t use the tool.7
Amazon’s closed-system approach contrasts with Shopify’s, which relies on open protocols, a structured data catalog of more than a billion products with clean attributes, real-time pricing, and inventory that is tailored to every major AI channel. Notably, Shopify reports that AI searches hitting that structured data catalog convert at roughly twice the rate of AI searches relying on scraped or stale data.
For AI users, AI search is three times more influential than traditional search
Globally, AI users are 3.2 times more likely to rank AI search tools as more influential than traditional search engines, and 2.6 times more likely to rank them higher than other strong influences, like friends or family (Exhibit 6).
Nearly a third of AI shoppers said AI chat or search tools were the biggest influence on their recent purchase.
1Question: Which sources had the biggest influence on your final decision? Top-ranked source. AI shoppers who described their most recent purchase, 11 markets, weighted, n = 5,095.
2Eg, Wirecutter.
That explains why AI users are almost twice as likely to use AI-based search first when researching or making a decision about what to buy as opposed to traditional search engines or other online sources of information. Responding to this trend, many traditional search engines have incorporated AI search into their capabilities.
When asked what actually drives AI users to use AI search while shopping, the top answer is saving effort. Eighty-one percent say AI does a good job narrowing down and filtering their choices; 80 percent cite less scrolling and clicking; and 79 percent cite significant time saved (Exhibit 7). Consumers are essentially delegating more of the research and evaluation work to AI agents.
Consumers said AI tools help them narrow down choices, scroll and click less, and save time.
1Question: When using AI tools while shopping, how much do you agree with the following statements? Top 2 box on a 5-point scale. AI shoppers, 11 markets, weighted, n = 9,645. Response not shown: “gets my search results right on the first try,” 67%.
And once they use AI, AI users are hooked. Nearly a quarter of AI users are now on an AI tool daily, and about one in ten use three or more AI tools daily.
Agentic commerce is helping e-commerce laggards make up a lot of ground
Agentic commerce is moving fast in areas that have traditionally lagged in e-commerce. As expected, some of the consumer categories pulling ahead in AI are typical e-commerce leaders (for example, apparel and travel). The surprise, however, is that sectors with traditionally low e-commerce penetration (such as grocery and beauty) are catching up quickly through agentic commerce (Exhibit 8).
In categories such as grocery and beauty, US e-commerce penetration is lower, yet more US consumers used AI to purchase in those categories.
1Question: Thinking about the past 3 months, how have you used AI tools or assistants (eg, ChatGPT, Gemini, Perplexity) when making purchases in the following categories? US AI users, weighted, n = 987. Grocery is 1 survey category (food, beverages, snacks, meal kits), so both grocery rows show the same figure.
One possible reason for this shift is that e-commerce grew to serve considered, comparison-heavy, ship-anywhere purchases but struggled to scale in categories defined by freshness, order complexity, and difficulty of returns. Those are precisely the categories where an agent that already knows the consumer’s household, shopping history, sizes, and standing preferences can remove the most friction.
EMARKETER’s modeling8 points to significant potential value: Widespread adoption of agentic commerce could increase US e-commerce sales by 11.5 percent above the company’s baseline forecasts by 2030.
Six priorities
As CMOs consider how to act on the emerging trends in agentic commerce, several priorities stand out:
-
Win the AI consideration set. Given the growing preference for using AI, CMOs should shift resources to maximizing the likelihood that their brands make an agent’s consideration shortlist. That includes investing in machine-readable content and product data, third-party authority, and AI-search optimization.
More broadly, CMOs should reconsider budgets to go after three groups of shoppers: a shrinking set of non-AI users to defend as cheaply as possible; a fast-growing set of samplers trying AI to win early; and a core of deep users whose reliance on AI is intensifying and whose habits are getting stickier to protect fiercely by becoming their default choice.
- Measure and optimize 24/7. Gathering insights on consumer behavior has long been a critical capability, but companies now need to account for consumers’ shopping agents as well. Companies need to understand how agents interpret consumer wants, which attributes and sources drive their recommendations, and why brands make, or miss, the consideration shortlist. And they need to measure continuously because agents don’t rest. In our experience, companies can put this structure and measurement in place within weeks and begin seeing impact within a few months.
- Make your product information easy for AI to use. AI agents can only recommend and buy your products if they have accurate, up-to-date information about them (such as specifications, prices, availability, delivery options, policies, and loyalty benefits). AI searches connected to a retailer’s product catalog convert at roughly twice the rate of searches relying on scraped or outdated information. Here’s a simple test: If an agent asked what a product costs, whether it is in stock, when it could be delivered, and whether it could buy it now, could your systems accurately respond? If not, improving the underlying product data and systems should be a priority.
- Protect your most critical assets. As agentic commerce grows, retailers will increasingly interact with a complex ecosystem of AI platforms, providers, agencies, commerce partners, and third-party AI tools. That makes it critical to decide what data and capabilities each partner can access, and on what terms. Product information, pricing, and inventory may need to be widely available for agents to recommend and transact effectively. But customer information, purchase and behavioral history, loyalty data, and proprietary insights are valuable assets that require greater protection. Retailers should establish clear rules for who can access this information, how it can be used and shared across partners, and what data third parties can retain or build from customer interactions.
- Invest in developing AI shopping assistants for owned sites. Retailer AI assistants may have less usage today, but their usage at the point of purchase indicates their growing potential value, especially when integrated with customer history, inventory, and fulfillment capabilities. Prioritize AI experiences that remove real shopping friction, such as helping customers narrow choices, compare products, find the right fit, or complete routine purchases.
- Put leaders in charge and give them sufficient authority. Agentic commerce allows for a much broader range of activity than e-commerce and requires greater speed. That makes it more important to bring together different functions, from marketing and merchandising to finance and legal, and to be clear on governance so decisions can happen quickly. That level of oversight and coordination requires a senior owner with authority to set platform and data rules, resolve trade-offs, manage risk, and move resources toward the categories and use cases showing the greatest potential. Much of this can and should be automated, with human oversight.
Consumers are clear they want to do more with AI. The question for retailers and brands is whether they are ready to capture that demand. Doing so will require rethinking how they reach and influence consumers, where they invest, and how they protect customer relationships. Those that figure out how to strike that balance can turn agentic commerce from a source of disruption into a new engine of growth.
