AI shopping statistics 2026: traffic is up 393%, but most stores can't be read

AI shopping stopped being a forecast and became a channel. AI-referred traffic to US retail sites grew 393% year over year in the first quarter of 2026, and by March those shoppers were converting better than visitors from regular search, according to Adobe Analytics. At the same time, independent studies keep finding that most stores leave the exact facts an AI assistant needs out of their markup. Demand and readiness are moving in opposite directions, and that gap is the opportunity.

This page collects the 2026 numbers we trust on AI shopping, AI search and store readiness, with the source named next to each figure. We keep it current and update it as new data lands. These figures come from different studies with different samples and methods, so read them as directional, not precise.

The short version

• AI store traffic is surging. AI-referred visits to US retailers were up 393% year over year in Q1 2026, and up 693% over the 2025 holiday season (Adobe Analytics).
• AI shoppers convert. In March 2026, AI-referred visitors converted 42% better than non-AI traffic, a reversal from 38% worse a year earlier (Adobe Analytics).
• Search is becoming answers. Google AI Overviews now appear on roughly 48% to 60% of queries depending on the tracker, and about 60% of searches end without a click.
• Most stores are not ready. Only 57.5% of top-ranking ecommerce pages used any schema at all in one 180-site study (Digital Chakra), and in our own check of 49 stores, more than 8 in 10 hid the machine-readable rating an assistant relies on.

How many people actually shop with AI now?

The honest answer is: a fast-growing minority, and they are valuable. In an Adobe survey of more than 5,000 consumers, 39% said they had used AI for online shopping, and 85% of that group said it improved their experience. That is adoption, not yet majority behaviour, but the trend in the traffic data is steep.

Adobe Analytics, which measures visits across a large set of US retail sites, reported:

Two things stand out. First, the conversion line flipped in a single year, from 38% worse to 42% better. The likely reason is that an assistant often does the comparison shopping first and only sends a visitor once it has narrowed the choice, so AI-referred shoppers arrive closer to buying. Second, Salesforce estimated that AI agents influenced more than 20% of all global online retail sales over the 2025 holiday season, which says the effect reaches well beyond the visits that show up as a direct AI referral.

How AI search is reshaping the storefront

Even when shoppers start on Google, the page they land on increasingly answers them in place. Google AI Overviews now appear on a large share of queries, though the exact figure depends heavily on the keyword sample and the tracker:

• BrightEdge put AI Overview coverage near 48% of tracked queries in early 2026.
• Advanced Web Ranking reported just over 60% of US queries.
• Conductor, on a different 21.9-million-query benchmark, measured about 25%.

The ranges disagree because the methods do, so we cite all three rather than pick the most dramatic one. What they agree on is the direction, and on the consequence: roughly 60% of Google searches now end without a click. For a store, that reframes the goal. The win is not only to rank, it is to be the source the answer is built from. A vendor benchmark from Prerender reported that around 66% of AI Overview citations come from sources that do not rank in the traditional top 10, which means classic ranking and AI citation are no longer the same race. (Prerender is a vendor in this space, so treat its benchmark as directional.)

How AI assistants decide which products to recommend

An AI shopping assistant does not look at your store the way a person does. It fetches the page, reads the machine-readable facts, and recommends what it can verify. Industry surveys converge on three factors that decide whether your product makes the shortlist:

1. Structured product data: your price, availability and rating present as data (Product schema / JSON-LD) the assistant can parse, not just visible on the page.
2. Citation authority: third-party signals like mentions on "best of" lists, reviews and references from other sites, which act like the AI version of domain authority.
3. Content quality and clarity: pages that state the answer plainly and consistently.

One AI-platform survey reported by Hexagon found that structured data, citation authority and content quality were named as the top three recommendation factors by the platforms it sampled. That is a vendor survey, so weigh it accordingly, but it lines up with what the readiness studies below keep showing: the assistant rewards the facts it can read, and skips the ones it cannot.

Are stores actually ready? The structured-data gap

This is where the opportunity sits, because the answer is mostly no, even for stores that look fine.

A Digital Chakra study analysed 180 top-ranking ecommerce pages across 12 niches in UK Google results and found only 57.5% used any schema markup at all. So roughly four in ten of the pages that already rank well were carrying no structured data, and "some schema" is a low bar that does not guarantee a complete, AI-readable Product block.

Our own first-party check points the same way. We ran 49 stores, a mix of well-known DTC brands and smaller independent shops, through the same read an AI shopping assistant performs. The result:

The stores were not broken. They scored 84 out of 100 on average, every one used HTTPS, and every one showed a price a shopper could read. They simply left specific facts out of the layer an assistant relies on, and the most common gap by far was the machine-readable rating: the review score written as data, not as star graphics. Forty-nine stores is a sample, not a census, and we treat it that way, but the pattern matched the third-party studies. Full method and raw data are in our 49-store AI shopping visibility study.

What the numbers add up to

Put the two halves together and the story is simple. The audience arriving from AI assistants is growing fast and converting better than search, while most stores still cannot be fully read by the assistants doing the recommending. The fix is rarely a rebuild. It is usually a few machine-readable facts, the product rating first, then price and availability, added to the markup an assistant already tries to read.

If you want to see which of those facts your own store exposes today, you can check it by hand: open a product page, view the source, and search for your price and for application/ld+json with a Product type. If your rating, price or availability are not in the markup, those are the gaps, and Google's Rich Results Test will validate any Product schema you add. If you would rather see every signal scored in one pass, ShelfGrader grades a store page free in about a minute.

Methodology and sources

We only include figures we could trace to a named source, and we label the source next to each number above. Where a figure comes from a vendor in the AI-visibility space (Prerender, Hexagon), we say so, because vendors have an interest in the trend. Where a figure is our own, we link the raw data. We do not invent search volumes or blend numbers from different studies into a single fake statistic.

• AI retail traffic, conversion, engagement and the consumer survey: Adobe Analytics, reported June 2026.
• AI-influenced share of holiday retail sales: Salesforce (2025 holiday season).
• Ecommerce schema adoption across 180 top-ranking pages: Digital Chakra schema markup study.
• AI Overview citation patterns and ecommerce AI coverage: Prerender AI indexing benchmark (vendor report, directional).
• AI Overview query coverage: BrightEdge, Advanced Web Ranking and Conductor (2026 trackers; figures vary by sample).
• Recommendation factors survey: Hexagon AI Platform Survey (vendor survey).
• The 49-store readiness numbers: our own AI shopping visibility study (first-party, 49 stores).

Last reviewed June 2026. We refresh this page as new and better data becomes available, so the numbers reflect what we could verify at that date, not a permanent claim.