Is the AI Shopping Market growing now?
In our AI shopping market deck, you will find everything you need to understand the market
SUMMARY
Yes. The AI shopping market is growing now, with the strongest growth happening in product discovery, comparison and retailer-owned shopping assistants rather than fully autonomous purchasing.
The clearest change is that AI traffic is surviving the move from a tiny base to a real ecommerce channel. Adobe’s AI-referred retail traffic growth has fallen from thousands of percent to triple digits, but it was still up 125% year over year between April and June 2026.
The quality of that traffic has changed even faster. AI-referred shoppers went from converting materially worse than other visitors in early 2025 to converting 42% better by Q1 2026, while generating 37% more revenue per visit.
Shopify’s merchant data makes the same point from another angle. AI-referred sessions grew more than eightfold year over year, but orders grew nearly thirteenfold, suggesting AI is getting better at sending shoppers who have already narrowed down what they want.
The market is meaningful without being dominant. General-purpose AI platforms are expected to drive more than $20 billion of U.S. ecommerce in 2026, but that is still only around 1.5% of total online retail sales.
Retailer-owned assistants are further along than the autonomous-agent narrative suggests. Amazon has already taken its shopping assistant to hundreds of millions of customers, while Walmart reports larger baskets among Sparky users and is putting the assistant deeper into its core shopping experience.
AI is gaining influence earlier in the purchase journey faster than it is taking over checkout. Consumers increasingly use assistants to research and compare products, yet 69% of AI-assisted shoppers still move to a retailer site or app to finish buying and only around 10% usually complete the purchase on the AI platform itself.
That leaves retailers in a surprisingly strong position. AI platforms can increasingly decide which two or three products a shopper considers, but retailers still control most of the inventory, loyalty, payment, delivery and returns infrastructure required to turn that recommendation into an order.
The infrastructure build-out is now becoming a market in its own right. Shopify, Google, OpenAI, Visa and Mastercard are working on product feeds, identity, carts, agent authentication and payment authorization because AI shopping is starting to create real operational problems rather than hypothetical ones.
Independent AI shopping startups still have room, but the generic shopping chatbot is already being squeezed. The better opportunities are appearing in categories where specialist product knowledge, resale data, fit, aesthetics or complex comparisons give a focused product an advantage over a general assistant.
Overall, AI shopping is growing much faster as an assisted-commerce channel than as a fully agentic one. Consumers are already comfortable letting AI help choose what to buy; letting an autonomous agent spend money and manage the whole transaction is developing much more slowly.
This market map, featured in our AI shopping market deck, highlights top companies and startups in the AI shopping market
What should we actually count as the AI shopping market?
The AI shopping market today covers the tools that help people discover, compare, choose or buy products through generative AI.
That includes general assistants such as ChatGPT, Gemini and Copilot, retailer-owned assistants such as Amazon’s Alexa for Shopping and Walmart’s Sparky, and specialist shopping products such as Phia, Daydream and Onton. It also includes the commerce infrastructure that lets these systems retrieve current product data or complete a transaction.
We would draw the line before ordinary retail AI. A forecasting model deciding how many shoes a retailer should stock clearly uses artificial intelligence, but the shopper never interacts with it. The market we are studying begins when AI becomes part of the customer’s actual path to purchase.
Market estimates get messy at that boundary. EMARKETER’s forecast for purchases driven by general-purpose AI platforms excludes Amazon’s and Walmart’s own shopping assistants. Broader studies from companies such as Salesforce include AI recommendations and conversational commerce inside retailers. Both can describe AI shopping accurately while producing completely different market sizes.
| What we count | Examples | Where it stands today |
|---|---|---|
| AI product discovery | ChatGPT, Gemini, Copilot | Already meaningful |
| Retailer AI assistants | Amazon Alexa for Shopping, Walmart Sparky | Operating at very large scale |
| Specialist AI shopping apps | Phia, Daydream, Onton | Smaller but growing |
| AI-assisted or agentic checkout | ChatGPT, Google, Copilot | Live but still early |
Is AI shopping traffic still growing now?
AI shopping traffic is still growing quickly today, although the crazy four-digit growth rates are naturally disappearing as the channel gets bigger.
Adobe Digital Insights has one of the best datasets for tracking this because it covers more than one trillion visits to U.S. retail websites. AI-referred retail traffic was up roughly 4,700% year over year in July 2025. During the 2025 holiday period, growth was still 693%. In the first quarter of 2026, AI referrals increased another 393% year over year.
More recent Adobe data shows the pace moderating further. Between April and June 2026, traffic coming from AI sources to U.S. retail websites was 125% higher than during the same period a year earlier.
The slowdown is useful. A channel cannot keep growing 4,700% once the comparison base stops being tiny. What we want to know now is whether growth survives that transition. So far, it has. AI shopping referrals have gone from thousands-of-percent growth to hundreds-of-percent growth while the underlying volume becomes much larger.
The pattern has also lasted long enough to rule out a one-off ChatGPT curiosity spike. We now have multiple holiday periods, multiple quarters and several independent commerce datasets pointing in the same direction.
| Period | Growth in AI-referred U.S. retail traffic |
|---|---|
| July 2025 vs. prior year | +4,700% |
| 2025 holiday period vs. prior year | +693% |
| Q1 2026 vs. prior year | +393% |
| April-June 2026 vs. prior year | +125% |
As this chart shows, and as featured in our AI shopping market deck, search interest in AI shopping has grown significantly
Are AI shoppers actually buying more products?
AI-referred shoppers are currently turning into unusually valuable ecommerce visitors, which is much stronger evidence of a real market than traffic growth alone.
Adobe found that AI-referred retail visitors converted 38% worse than other traffic sources around the beginning of 2025. By the first quarter of 2026, the relationship had completely flipped: they were converting 42% better and generating 37% more revenue per visit.
Shopify sees an even clearer pattern inside actual merchant transactions. Its Q1 2026 commerce data shows referral sessions from AI chatbots growing more than eightfold year over year while AI-referred orders grew nearly thirteenfold. Orders therefore grew much faster than visits.
Shopify also compared AI referrals directly with organic search. On sessions beginning on product pages, AI-referred shoppers converted nearly 50% better and generated 14% higher average order values. AI beat organic search on conversion across 23 of the 25 merchant categories Shopify examined.
A likely reason is that someone coming from an AI assistant has often already explained what they need, compared choices and narrowed the decision before reaching the store. The retailer receives a visitor who is further down the buying journey.
| Shopify Q1 2026 metric | AI shopping performance |
|---|---|
| Referral-session growth | More than 8x YoY |
| AI-referred order growth | Nearly 13x YoY |
| Conversion vs. organic search | Nearly 50% higher |
| Average order value vs. organic search | 14% higher |
| Categories where AI beat organic conversion | 23 of 25 |
How big is the AI shopping market today?
The AI shopping market is already worth tens of billions of dollars in the U.S., but it still represents a small share of total ecommerce.
EMARKETER expects purchases driven through general-purpose AI platforms such as ChatGPT, Gemini and Perplexity to exceed $20 billion in U.S. retail ecommerce this year. That works out to roughly 1.5% of total online retail sales.
The same forecast has AI-platform-driven ecommerce rising above $144 billion by 2029, or 8.8% of U.S. retail ecommerce. Even the current figure therefore implies a channel moving from roughly niche scale toward something large enough for every major retailer to care about.
The 1.5% figure also keeps the story grounded. Most online purchases still happen without a general AI platform driving the transaction, and retailer websites remain the place where most AI-assisted shoppers eventually pay.
The broader economic influence of AI shopping is already much larger because the $20 billion estimate deliberately leaves out retailer-native assistants such as Amazon’s and Walmart’s. Salesforce found during Cyber Week 2025 that AI and agents influenced about $67 billion of global spending, or roughly 20% of purchases, using a much broader definition that included personalized recommendations and conversational assistance.
So there are really two scales developing at once: direct AI-platform commerce remains relatively small, while AI-assisted buying decisions have become much more common.
This chart, featured in our AI shopping market deck, illustrates yearly VC funding for AI shopping startups
Is Amazon proving that AI shopping works at mass scale?
Amazon currently provides the strongest evidence that an AI shopping assistant can work with hundreds of millions of ordinary consumers.
Amazon said Rufus, now folded into Alexa for Shopping, was used by more than 300 million customers during 2025 and helped generate nearly $12 billion in incremental annualized sales. In Amazon’s latest quarterly update, the company said more than 350 million customers had used its shopping assistant over the previous twelve months.
Usage is still climbing sharply. Amazon reported that active users were close to doubling year over year while interactions increased more than fivefold. Earlier company data also showed that customers using the assistant during a shopping trip were more than 60% more likely to make a purchase.
Alexa for Shopping now goes much further than answering questions. It compares products, uses price history, tracks prices and can automatically buy an item when it reaches a price chosen by the customer. Amazon has also experimented with buying products from outside its own marketplace.
One caveat: people with stronger purchase intent may simply be more likely to use the assistant in the first place, so the 60% purchase gap cannot all be credited to AI. The scale is harder to argue with. Hundreds of millions of customers are already interacting with a generative shopping interface inside one of the world’s largest ecommerce businesses, and engagement is still accelerating.
If you want more recent data on this point, please see our latest AI shopping market report.
Is Walmart seeing the same AI shopping behavior as Amazon?
Walmart is seeing the same basic pattern: customers who use its AI shopping assistant build bigger baskets, and the company keeps putting Sparky deeper into the shopping experience.
During Walmart’s fiscal-year-end earnings call, management said Sparky users had average order values roughly 35% higher than customers who did not use Sparky. In a follow-up investor call, Walmart added that roughly half of customers engaging with its app were also engaging with Sparky.
Walmart has been expanding what the assistant can do rather than leaving it as a search feature. Sparky can help build baskets, answer product questions and connect a shopper’s intent directly with Walmart’s delivery, pickup and store network.
The company is also taking Sparky outside its own app. OpenAI’s latest commerce update introduced a Walmart experience inside ChatGPT that connects conversational discovery with Walmart account linking, loyalty and payments.
That combination is more interesting than another chatbot launch. Walmart is trying to join AI discovery wherever it happens while keeping its retail identity, customer account and transaction system attached to the purchase.
Amazon and Walmart are approaching the problem differently, but both have now found enough engagement to keep expanding their shopping assistants. Retailer-native AI is already one of the most developed parts of this market.
This chart, featured in our AI shopping market deck, shows why Constructor is winning in AI shopping
Are people actually changing how they discover products?
Consumers are already using AI to decide what to buy, especially when the purchase requires comparison rather than a simple keyword search.
Adobe consumer research found that people were using generative AI for tasks including product research, recommendations, deal hunting and finding unusual products. Among AI-shopping users, research was the most common activity.
The behavioral data fits that use case. AI-referred shoppers arrive at ecommerce sites with unusually high conversion rates because much of the messy research has already happened inside the conversation. Shopify also found that more than half of AI-referred product-page sessions started directly on the product page, compared with roughly one-fifth for organic search.
It is a different shopping journey. Someone searching Google for “running shoes” may still click several category pages and comparison sites. Someone telling an AI assistant, “I run 30 kilometers a week, have wide feet and want something under $150” can arrive directly at two or three plausible products.
AI shopping looks strongest these days when the customer knows the problem but does not yet know the product. That covers fashion, electronics, beauty, home products, gifts and other categories where shoppers normally spend time comparing options.
Traditional search still handles an enormous amount of ecommerce discovery. What has changed is that conversational AI now owns part of the consideration process that previously required several searches, tabs and retailer visits.
Is AI shopping actually replacing Google Search?
AI shopping is changing how people use Google more quickly than it is replacing Google itself.
Google has responded by turning its own shopping surfaces into AI products. Its Shopping Graph now contains more than 60 billion product listings, and Google says people perform more than one billion shopping interactions across its services each day.
Gemini and AI Mode can use that catalog to compare products conversationally rather than returning only a list of links. Google has also introduced Universal Cart, which lets a shopper add products while using Search or Gemini and eventually across other Google services.
Google is pushing further into the transaction itself. Its Universal Commerce Protocol can retrieve current pricing and inventory, keep loyalty benefits attached to the customer and support checkout on Google for participating retailers. Universal Cart can also track deals, price drops and stock availability.
Google has an unusual advantage here. It already owns enormous shopping-intent distribution, a vast merchant catalog, advertising relationships and payment infrastructure. Generative AI can be inserted into that existing machine.
The competitive question has shifted. ChatGPT and other assistants can take product-discovery traffic away from traditional Google results, but Google is simultaneously rebuilding Search around the same conversational behavior. Classic search pages can lose importance without Google losing its role in commerce to the same degree.
If you want more recent data on this point, please see our latest AI shopping market report.
This chart, featured in our AI shopping market deck, illustrates yearly funding for AI shopping startups
Are ChatGPT, Gemini and Copilot becoming real shopping channels?
ChatGPT, Gemini and Copilot are becoming real shopping channels now because major merchants can increasingly expose products to them without building one-off integrations.
OpenAI has made product discovery a dedicated part of ChatGPT. Shoppers can browse products visually, compare them side by side and refine recommendations through conversation. OpenAI also says product information from Shopify merchants is now integrated through Shopify Catalog, while retailers including Target, Sephora, Nordstrom, Lowe’s, Best Buy, Home Depot and Wayfair have connected product discovery through its commerce protocol.
Google is doing something similar across AI Mode and Gemini. Its Universal Commerce Protocol was built with companies including Shopify, Etsy, Wayfair, Target and Walmart and has support from a much larger group of retailers and payment companies.
Microsoft Copilot can also surface Shopify products and support checkout for eligible merchants.
The real change is distribution. A small brand previously needed to optimize for Google, marketplaces, social platforms and perhaps retailer search. Millions of Shopify merchants can now appear across several AI interfaces through infrastructure handled by Shopify itself.
For merchants, AI assistants are starting to look less like experimental chatbots and more like another place where products need to be discoverable.
Is AI checkout actually happening today?
AI checkout is live today, but product discovery is clearly further ahead than letting an agent complete purchases inside the conversation.
The latest OpenAI changes make this especially clear. OpenAI launched Instant Checkout as an early attempt to keep the whole purchase inside ChatGPT, but the company later said the first version did not give merchants enough flexibility. Its newer shopping strategy puts more emphasis on product discovery and lets merchants use their existing checkout experiences.
ChatGPT still supports Instant Checkout for some eligible products and merchants, but Shopify merchants generally send the shopper through the merchant’s own checkout, including through an in-app browser. Google similarly lets participating shoppers either buy through its UCP-powered checkout or move the cart to the retailer.
Consumer behavior looks similar. According to a Publicis Commerce and EMARKETER survey, only about 10% of AI-assisted digital shoppers usually complete the purchase on the AI platform itself. Sixty-nine percent move to a retailer’s website or app.
That gap is one of the clearest markers of where AI shopping currently stands. People are comfortable asking AI what to buy. They are less willing to hand the entire transaction to the assistant.
The autonomous version of agentic commerce therefore has a long way to go, even though AI-assisted commerce is already growing quickly.
This chart, featured in our AI shopping market deck, compares the main business model options for AI shopping assistants
Are retailers taking AI shopping seriously yet?
Retailers are taking AI shopping seriously now because measurable orders are arriving from the channel and the technical cost of participating has fallen sharply.
Shopify’s commerce data is hard to ignore. As seen above, AI-referred sessions grew more than eightfold year over year in Q1 2026 while AI-referred orders increased nearly thirteenfold. The same dataset found that AI referrals converted better than organic search in 23 of 25 merchant categories.
Shopify has responded by making Agentic Storefronts available across millions of merchants. Products can appear in ChatGPT, Microsoft Copilot, Google AI Mode and Gemini from the same commerce backend. Orders are attributed back to the AI channel, while merchants keep control of customer relationships and remain merchant of record.
The infrastructure is also becoming much more practical. Shopify Catalog structures billions of products for AI discovery. Google’s Universal Commerce Protocol handles things such as live product information, carts, identity and checkout. OpenAI’s Agentic Commerce Protocol provides another route for merchants to feed current product information into AI experiences.
Retailers have moved past simply asking whether consumers might eventually shop through AI. They are now dealing with product feeds, attribution, loyalty, inventory and checkout integration. Those are the problems that appear once a channel starts producing actual commerce.
Why are Visa and Mastercard building specifically for AI shopping agents?
Visa and Mastercard are building dedicated infrastructure because AI shopping agents create payment and identity problems that normal ecommerce checkout was never designed to handle.
A retailer needs to know whether an automated visitor is a legitimate shopping agent or a malicious bot. It also needs to know which customer authorized that agent, what the agent is allowed to do and whether the payment credential is valid for that specific purchase.
Visa’s Trusted Agent Protocol tackles exactly those questions. It allows approved agents to identify themselves cryptographically and communicate their shopping intent to merchants. Visa has also built Intelligent Commerce products aimed at connecting agents, merchants and payments.
Mastercard is pursuing the same area through Agent Pay. Its newer Agent Pay for Machines initiative has support from more than 30 companies across payments, infrastructure and technology, including Adyen, Checkout.com, Cloudflare, Global Payments and Stripe.
Google has developed its own Agent Payments Protocol, while PayPal is integrating its wallet into AI-assisted shopping experiences.
Payment companies would obviously like this market to happen, so their investment does not prove large transaction volumes by itself. But the industry is already solving agent authentication, payment authorization and fraud rather than debating whether an AI can recommend a pair of shoes. That is a pretty meaningful shift.
If you want more recent data on this point, please see our latest AI shopping market report.
This chart, featured in our AI shopping market deck, shows how market revenue is split across customer segments in the AI shopping market
Are AI shopping startups getting real traction or just raising money?
Some AI shopping startups are showing real usage and revenue growth, although the category is still young enough that a few fast-growing companies should not be confused with a mature startup market.
Phia is the clearest recent example. After launching in 2025, the fashion shopping agent raised an $8 million seed round and then a $35 million Series A only months later. The company said it had passed one million users, connected with more than 6,200 retail brands and grown revenue elevenfold since launch. Its Series A valued the business at $185 million.
Onton provides a second useful example outside fashion price comparison. The company said its monthly active users grew from about 50,000 to more than two million before it raised another $7.5 million to expand beyond furniture.
Daydream took the opposite route: enormous funding before large-scale consumer proof. The company raised a $50 million seed round to build conversational fashion discovery and later launched its product publicly after assembling a catalog across thousands of brands.
These companies are testing different ways to capture the same behavioral shift. Phia focuses heavily on price comparison and resale alternatives. Daydream tries to understand vague fashion intent. Onton built around visual and conversational product discovery.
The traction is real enough to keep the category interesting. What remains unclear is whether independent shopping apps can own a large consumer relationship when Amazon, Google, OpenAI and Shopify are building similar capabilities directly into products people already use.
| AI shopping startup | Disclosed funding | Reported traction |
|---|---|---|
| Phia | More than $43M | 1M+ users, 6,200+ retail brands, 11x revenue growth |
| Daydream | $50M seed | Conversational fashion platform with thousands of brands |
| Onton | $7.5M latest round | MAUs grew from ~50K to 2M+ |
Can AI shopping startups survive against Amazon, Google and ChatGPT?
AI shopping startups can still build large businesses, but a generic “ChatGPT for shopping” has become much harder to defend.
Amazon has purchase history, current inventory, delivery information and hundreds of millions of existing shoppers. Google brings more than 60 billion product listings and massive search distribution. ChatGPT already has a huge general-purpose user base. Shopify can make millions of merchants accessible to multiple AI interfaces at once.
A startup therefore needs something those platforms cannot reproduce cheaply. Phia focuses on comparing new and resale inventory across fashion retailers. Daydream is trying to understand the language and visual nuance of fashion rather than treating every catalog as generic products. Onton built deeper product discovery around categories where specifications, dimensions and aesthetics can make ordinary keyword search frustrating.
Distribution matters just as much as model quality. An independent shopping agent must convince consumers to open another app or install another extension, then somehow monetize the purchase. Many currently rely on affiliate commissions, which makes attribution and retailer relationships critical.
There is still room for specialists because shopping is too broad for one generic interface to be best at everything. Fashion fit, luxury resale, furniture, beauty ingredients and electronics comparison each require different product knowledge.
Still, the bar has risen quickly. A startup that merely turns product search into a chat interface is competing with features that the largest commerce and AI companies can now ship themselves.
If you want more recent data on this point, please see our latest AI shopping market report.
This chart, featured in our AI shopping market deck, shows how AI shopping assistant technology has evolved over time
Who is actually winning AI shopping right now: retailers or AI platforms?
Retailers are currently winning the transaction layer of AI shopping, while general AI platforms are gaining power earlier in the decision.
EMARKETER’s latest forecast estimates that retailer-native AI assistants will drive 54.1% of U.S. AI-driven retail ecommerce sales this year. The firm expects retailers to remain ahead of general-purpose AI platforms for several years.
The consumer journey explains why. Publicis Commerce and EMARKETER found that 69% of AI-assisted digital shoppers normally go to a retailer’s site or app to finish buying, compared with only 10% who usually complete the purchase on the AI platform.
Trust also favors retailers. Bain research cited by EMARKETER found that 25% of U.S. shoppers trusted retailers most to manage the complete shopping experience, compared with 16% for technology companies such as Google and only 7% for AI platforms such as ChatGPT.
Amazon and Walmart have another advantage: they know what the customer bought before, what is in stock, what can arrive tomorrow, which loyalty benefits apply and how a return should be handled.
General AI platforms can still become extremely powerful gatekeepers. If ChatGPT or Gemini reduces 200 possible products to three recommendations before the shopper reaches Amazon, a large part of the commercial decision has already happened.
Right now, retailers own more of the purchase while AI platforms are fighting to own more of the choice.
| Current AI-shopping advantage | Retailers | General AI platforms |
|---|---|---|
| Share of AI-driven ecommerce led by retailer-native assistants | 54.1% | Minority share |
| Shoppers usually completing transaction there | 69% | 10% |
| Consumers naming them most trusted for full journey | 25% | 7% for AI platforms |
Why aren't people letting AI agents buy everything for them yet?
Trust, merchant control and transaction complexity are currently keeping fully autonomous AI shopping well behind AI-assisted discovery.
Buying a product involves more than choosing the right item. The agent may need current inventory, shipping restrictions, loyalty status, payment authorization, return rules, tax information and confidence that the merchant will recognize the transaction as legitimate.
Consumers also have a different tolerance for mistakes once money moves. A poor product recommendation wastes a few minutes. An agent ordering the wrong size, paying the wrong price or buying from an unreliable merchant creates an actual financial problem.
Retailers have their own concerns. They want AI platforms to send customers without taking ownership of pricing, payments, customer data and post-purchase relationships. OpenAI’s decision to give merchants more checkout flexibility reflects that tension directly.
The industry is working through these problems now. Google’s commerce protocol supports identity linking and loyalty. Visa’s protocol helps merchants distinguish authorized shopping agents from bots. Mastercard is building agent-specific payment infrastructure. Shopify keeps merchants as the seller of record across its AI channels.
Those developments should make autonomous purchasing easier, but consumer adoption will probably move one permission at a time. Asking AI to compare ten laptops is already normal for many users. Letting it independently spend $2,000 remains a much bigger step.
If you want more recent data on this point, please see our latest AI shopping market report.
In our AI shopping market deck, we identify pain points entrepreneurs should prioritize
Is the AI shopping market growing now?
Yes. The AI shopping market is clearly growing now, and the evidence has moved well beyond product announcements and hype.
We have several independent measurements pointing in the same direction. Adobe still sees triple-digit growth in AI retail referrals from a much larger base. Shopify sees AI-generated orders rising even faster than AI traffic, with conversion beating organic search across almost every merchant category it studied. Amazon has taken its shopping assistant to hundreds of millions of users while engagement keeps rising sharply. Walmart has found materially larger baskets among Sparky users and is now connecting its shopping experience directly into ChatGPT.
The infrastructure is expanding at the same time. Millions of Shopify merchants can participate in AI shopping channels. Google is connecting Search, Gemini, product catalogs and checkout. OpenAI has made richer shopping discovery a core ChatGPT experience. Visa, Mastercard and other payment companies are building specifically for transactions initiated by AI agents.
The market is still uneven. General-purpose AI platforms are expected to drive only around 1.5% of U.S. ecommerce this year, and most AI-assisted shoppers still leave the AI platform before completing the purchase. OpenAI’s own shift toward merchant-controlled checkout shows how early fully agentic buying remains.
If anything, that clarifies where the growth is actually happening.
AI shopping today is strongest in product discovery, comparison and recommendation. Retailer-native assistants have already reached meaningful scale. AI-referred traffic is becoming commercially valuable. Native and autonomous checkout is developing more slowly.
The strongest evidence is the change in buyer behavior rather than any single launch: AI shopping traffic keeps growing while the people arriving through that channel increasingly convert and spend like high-intent customers.
So the answer is yes, with one important boundary. The AI shopping market is growing fast today, while the fully autonomous “tell an agent what I need and let it buy everything for me” market is still early.
OUR METHODOLOGY
This analysis tests whether the AI shopping market is genuinely growing based on observed consumer behavior, commercial activity and infrastructure development. Because the market spans product discovery, recommendations, retailer assistants, referred commerce, checkout and increasingly autonomous transactions, we did not treat any single market-size estimate or product launch as the answer.
We broke the question into several analytical dimensions: consumer behavior, AI-referred traffic, conversion and orders, retailer adoption, platform distribution, transaction infrastructure and specialist AI-shopping activity. Each dimension answers a different part of the growth question, so we treated them as complementary rather than interchangeable measures.
We gave the most weight to observed shopping activity. Traffic data showed whether consumers were reaching retailers through AI, while conversion, orders, average order values and revenue per visit helped determine whether that traffic was becoming commercially useful rather than simply generating curiosity clicks.
Recency was important because the question is whether the market is growing now. We therefore prioritized 2026 data and the latest available 2025–2026 operating updates, using earlier figures mainly as baselines for measuring how traffic, conversion, engagement or adoption had changed.
We also separated direct AI-platform commerce from broader AI-assisted shopping. EMARKETER’s estimate of ecommerce driven by general-purpose AI platforms measures a narrower channel than Salesforce’s estimate of spending influenced by AI and agents. We use both, but not as if they were measuring the same market.
Retailer-native assistants were assessed separately because Amazon and Walmart already have customer accounts, product catalogs, inventory, loyalty programs and checkout infrastructure that general-purpose AI platforms do not fully control. Their usage, engagement and order-value data therefore provide evidence about a more mature part of AI shopping.
We treated fully autonomous purchasing as a later stage of the market rather than a requirement for growth. The analysis therefore distinguishes between AI helping a shopper discover and compare products, AI referring that shopper to a merchant, and an agent independently completing the transaction.
We did not use a mechanical scoring model. The underlying evidence measures different things and comes from datasets of very different scale. Instead, we looked for convergence across independent sources and gave more weight to evidence tied directly to shopping behavior, transactions and operating adoption than to forecasts, funding rounds or product announcements.
Key sources used for the analysis include Adobe Digital Insights on Q1 2026 AI retail traffic and conversion, Adobe on April–June 2026 AI-referred retail traffic, Shopify on AI-referred orders and conversion, Amazon’s Q4 results on shopping-assistant usage and incremental sales, Walmart’s FY26 earnings call on Sparky engagement and order value, EMARKETER on U.S. ecommerce sales driven by AI platforms, and EMARKETER on retailer-native AI, checkout behavior and consumer trust.
We also used Salesforce’s Cyber Week data on spending influenced by AI and agents, OpenAI’s product-discovery and retailer-integration documentation, OpenAI’s Instant Checkout and Agentic Commerce Protocol documentation, Google’s Universal Commerce Protocol announcement, Google’s Universal Cart and Shopping Graph update, Visa’s Trusted Agent Protocol documentation, Visa Intelligent Commerce, and Mastercard’s Agent Pay for Machines announcement.
This chart, featured in our AI shopping market deck, shows how market revenue is split across Europe, Asia, North America, Africa, and South America in the AI shopping market
Related blog posts
