AI Shoppers Now Outconvert Humans 42% on US Retail Sites

AI assistants are now sending US retailers shoppers who buy more often, browse more pages, and spend more per visit than the humans clicking in from Google. Adobe Analytics reported on April 16 that traffic from AI sources to US retail sites grew 393% year over year in the first quarter of 2026, and those visitors converted 42% better than non-AI traffic in March, a record reversal from a 38% deficit just twelve months earlier.

That flip inverts the year-old playbook that treated chatbot referrals as low-intent browsing. It also exposes a structural gap retailers can no longer postpone: roughly a quarter of US retail homepage content, and a third of average product page content, is still unreadable by the very models generating the demand.

The Conversion Curve Just Flipped

A year ago the picture was the opposite. In March 2025, visitors arriving from generative AI tools converted 38% worse than shoppers from paid search, email, or organic search. Twelve months later the same comparison runs the other way, and the average revenue per visit from AI referrals sits 37% above the rest of the channel mix.

The shift was not a gradual taper. December’s holiday window already showed AI traffic running 693% above the prior year, and the March-only figure of 269% suggests the post-holiday cooldown most analysts expected did not arrive. The Q1 sample covers more than one trillion visits to US retail sites, the full Adobe Analytics panel.

AI is quickly becoming the primary interface between consumers and their favorite brands.

That line is from Vivek Pandya, director of Adobe Digital Insights, in the same Q1 report. The framing matters because the conversion premium is the first quarter where it appears in monthly data, not a single anomaly week.

Why Shoppers Sent by an AI Behave Differently

Once a visitor lands on a retail page from an AI assistant, the on-site behavior is measurably stickier. They spend 48% more time on the page, browse 13% more pages per visit, and register a 12% higher engagement rate than visitors from any other channel.

One reading is selection bias. By the time an assistant routes a user to a specific product URL, the discovery, comparison, and category filtering have already happened inside the chat. The click that lands on the site is closer to a near-purchase intent than a top-of-funnel browse.

Trust is the second factor. Adobe surveyed more than 5,000 US consumers alongside the traffic data; 39% said they have used AI for online shopping, 85% of that group said it improved the experience, and 66% said they believe AI tools provide accurate results. Conversion does not climb if the recommendation feels like an ad.

The third factor is supply. Retailers with branded AI agents sitting on their own sites grew holiday sales 32% faster than those without, according to Salesforce’s 2025 holiday data. The conversion lift is partly a function of retailers meeting the AI traffic with infrastructure built for it.

A Quarter of Retail Content Is Invisible to the Models

The same Q1 report includes the first wide read from Adobe’s AI Content Visibility Checker, a scoring tool that measures how much of a page is parseable by large language models. A score of 50% means half the content on the page cannot be read by an LLM at all.

Homepages Lead, Help Pages Trail

The average US retail homepage scored 75%. Returns and exchanges pages scored 82%, contact pages 81%, and store locators came in at 73%. Category pages such as appliances or men’s apparel scored 74%, almost identical to homepages.

Those numbers sound healthy until they are compared to where shoppers actually decide.

Product Pages Carry the Largest Gap

Individual product pages averaged just 66%. That is the page where price, availability, sizing, return policy, and reviews live, and it is the page an AI agent must parse to make or recommend a purchase. A third of that information is currently dark to the models.

For retailers carrying tens of thousands of SKUs, the readability gap is not a one-off fix on a hero page. It is structural data hygiene applied across every product template.

Best vs Worst Performers

• 82.5%: average homepage score for the best-performing decile of US retailers
• 54.2%: average homepage score for the worst-performing decile
• 28-point spread: the readability gap between the two ends of the same channel

Adobe ships an LLM Optimizer Chrome extension that lets any retailer score its own pages against the same benchmark. The same Adobe team has been on this beat since 2024, and the gap between the best and worst sites is wider than the year-over-year improvement, which is the metric that should worry the back half of the table.

The Court Fight Over Who Owns the AI Checkout

The conversion data is landing into a legal and product environment that has not settled who gets to handle the checkout step. Three flashpoints from the past nine months frame the contest.

Amazon’s Win Against the Comet Browser

On March 9, US District Judge Maxine Chesney granted Amazon a preliminary injunction blocking Perplexity’s Comet browser from shopping on Amazon on behalf of users. The court found that Amazon was likely to win its claims under the federal Computer Fraud and Abuse Act and a California computer fraud statute, on the theory that Comet accessed accounts with user permission but without Amazon’s authorization.

Amazon said it warned Perplexity five times beginning in November 2024, deployed a technical block in August 2025, and watched Perplexity push a workaround inside 24 hours. The order requires Perplexity to destroy collected Amazon data, and Perplexity appealed on March 11. The same authorization question is reshaping how payment networks handle AI agent transactions.

OpenAI’s Short-Lived Instant Checkout

OpenAI launched Instant Checkout inside ChatGPT on September 29, 2025, built on a Stripe-codeveloped Agentic Commerce Protocol, starting with US Etsy sellers and a planned rollout to Shopify merchants. The feature was shut down within six months. OpenAI said the initial version did not offer the flexibility the company aspired to, and pointed to a coming multi-cart redesign.

The pull-back is the rare admission that the first generation of in-chat purchase flows did not match the conversion behavior consumers already showed on retailer-owned pages.

Salesforce’s Holiday Benchmark

Salesforce measured that AI and AI agents influenced 20% of all global online retail sales during the 2025 holiday season, alongside a 66% jump in agentic AI customer-service conversations in December and a 142% surge in AI-managed return and shipping actions. Global holiday online sales reached $1.29 trillion, $294 billion of it in the US. Meta and Google are racing to ship the consumer AI agents that would sit on the other side of that traffic, and OpenAI is rebuilding its own.

What Retailers Are Reaching For in Q2

The actionable read from the Q1 figures, for any retailer with a product catalog, breaks into a short list of moves that the better-performing decile already runs.

• Score the product templates first. The homepage average is 75%, but the page that decides a sale averages 66%. Closing the product-page gap is the highest-leverage fix in the building.
• Expose structured data, not marketing copy. AI assistants parse JSON-LD product schema, inventory feeds, and return policies. Hero animations and image-only specifications stay invisible.
• Stand up an on-site agent. Retailers with their own branded agents grew holiday sales 32% faster than those without, per Salesforce.
• Treat the AI referrer as a paid channel. Revenue per visit is now 37% above non-AI traffic; attribution and budget allocation need to follow the money.

McKinsey’s projection that agentic commerce will drive $1 trillion in US retail revenue by 2030 is the size of the prize, and it is the reason every layer of the stack (assistants, browsers, payment networks, platforms) is trying to claim a piece. If Q2 visibility scores climb past the current 66% product-page average before the holiday window opens, the conversion lead compounds. If they do not, AI traffic will keep growing and converting somewhere else, on the sites that did the work.

Frequently Asked Questions

What does Adobe’s Q1 2026 AI retail traffic report actually measure?

The report covers more than one trillion visits to US retail websites during January, February, and March of 2026, paired with a survey of over 5,000 US consumers. The traffic figures isolate visits referred from generative AI sources such as ChatGPT and other assistants, and the visibility figures come from Adobe’s AI Content Visibility Checker, which scores how much of a webpage is parseable by LLMs.

Why did AI shoppers shift from converting worse to better in twelve months?

Three factors stacked. Assistants improved at routing users to specific product URLs rather than category pages, so the click that lands on the site is closer to purchase intent. Consumer trust climbed: 66% of Adobe’s surveyed shoppers said AI tools provide accurate results. And retailers with branded on-site agents grew holiday sales 32% faster than those without them, indicating that part of the lift comes from sites meeting the AI traffic with infrastructure built for it.

How can a retailer check whether its own pages are readable by AI assistants?

Adobe ships a Chrome extension called AI Content Visibility Checker, powered by its LLM Optimizer, which scores any URL on a 0-to-100 scale. A 50% score means half the content on the page cannot be parsed by an LLM. The Q1 benchmarks are 75% for the average US retail homepage, 74% for category pages, and 66% for individual product pages.

Did the Amazon v Perplexity ruling stop AI agents from shopping?

No. The March 9 preliminary injunction is specific to Comet accessing Amazon. The court ruled that user permission alone is not enough; platform authorization is a separate legal requirement under the Computer Fraud and Abuse Act. Other retailers can still invite agentic traffic, and most appear to be doing so. The injunction is stayed pending Perplexity’s Ninth Circuit appeal filed March 11.

Is agentic commerce already a meaningful share of retail revenue?

Yes, by the Salesforce read. AI and agents influenced 20% of global online retail sales during the 2025 holiday season, the period when AI customer-service conversations rose 66% year over year and AI-managed return and shipping actions rose 142%. McKinsey projects US agentic commerce alone could orchestrate $1 trillion in retail revenue by 2030.

What is the single biggest readability fix a retailer can make in Q2?

Closing the product-page gap. The average homepage scores 75%, but the page where price, sizing, availability, reviews, and return policy live averages 66%. That is the page an AI agent must parse to make or recommend a purchase, and the highest-leverage fix is exposing structured product schema and return information as machine-readable data rather than image-only marketing copy.