State of AI Commerce 2026: Key Findings
We ran 58,320 tests to see which stores AI recommends when shoppers ask what to buy. The full data, open and free to cite.
The numbers, free to lift.
AI does not recommend the giants.
The intuitive assumption is that AI recommendations flow to the largest retailers. They do not. Across every loss in the study, AI recommended 50,287 different brands. The single most-recommended brand on the entire internet, Etsy, accounted for just 1.3% of recommendations. Walmart took 0.6%.
The hundred most-recommended brands combined made up only 11% of the total, and 71% of all winning brands were recommended exactly once. There is no incumbent that owns AI search. The recommendation goes to the store whose catalog most clearly answers the specific question asked.
On Google, a few incumbents dominate every commercial query. In AI recommendation, the most dominant brand on earth wins barely more than 1% of the time. The field is wide open.
State of AI Commerce 2026Free to embed with credit| Brand | Share of all AI recommendations |
|---|---|
| Etsy#1 | |
| Walmart | |
| Target | |
| Nordstrom | |
| eBay | |
| ASOS | |
| Anthropologie | |
| Amazon | |
| Patagonia | |
| Lululemon |
Six in ten stores are invisible to AI.
When each of 9,720 stores was tested with six purchase-intent queries, 60% were recommended in none of them. Across all 58,320 tests, stores were recommended 14% of the time, a competitor 79% of the time, and 7% of the time AI named no store at all.
Visibility is close to all-or-nothing: 60% of stores won nothing, only 0.8% won all six queries, and of the stores that won anything, 70% won just one or two. The middle is nearly empty.
State of AI Commerce 2026Free to embed with creditMost stores win nothing; almost none win consistently. The distribution splits cleanly in two.
Your category sets your odds before you do anything.
The single biggest predictor of whether a store gets recommended is its product category. Win rates run from 17% in sports, pets, and hobbies down to 9% in home and living. The share of invisible stores runs from 52% in food and beverage, the most visible category, to 74% in home and living, the least.
AI rewards categories where products have clear, describable specifications, and struggles in categories driven by visual taste, where it falls back on a few large generic retailers.
State of AI Commerce 2026Free to embed with credit State of AI Commerce 2026Free to embed with creditThe shoppers arriving through AI are the most valuable traffic a store gets.
The shift is measurable. AI-referred traffic to US retailers grew 393% in the first quarter of 2026 over the year before, according to Adobe Analytics. A year earlier that traffic converted 38% worse than a regular visitor; as of March 2026 it converts 42% better and produces 37% more revenue per visit.
The shoppers arriving through AI are now the most valuable traffic an online store gets — and this study shows most stores are not built to receive it.
The stores that win every query share one trait.
Seventy-eight stores were recommended for all six of their queries. They are overwhelmingly focused, single-category brands with one clear product identity — companies an AI engine can describe in a single sentence.
When a catalog has a sharp, consistent identity, the engine knows exactly which question it answers. When a catalog is sprawling or vague, it has nothing clean to recommend it for. Focus reads as authority to a machine the same way it does to a person.
58,320 head-to-head comparisons, every win backed by a proof page.
Recomaze scanned 9,720 ecommerce store domains sourced from BuiltWith, spanning global brands to small independent shops. Each store was tested with six purchase-intent queries matched to its product category, submitted to Google Gemini in April 2026. Each response was classified as a win (the store was recommended), a loss (a competitor was recommended), or no recommendation. That produced 58,320 head-to-head comparisons across 58,043 distinct queries. Every win is backed by a public proof page showing the exact query and the raw AI response.
scanned
queries per store
comparisons
in this dataset
Single scan per store, not an average of repeated runs. Google Gemini only. Queries were generated algorithmically, so some no-recommendation results reflect query phrasing rather than store quality. Product categories were assigned by classifying query language, with roughly 42% uncategorized.
These findings are free to cite.
The charts are free to embed with credit. For the full data, methodology, additional findings, or to request an interview, get in touch below. Copy what you need.
Delian Coroamă, Founder & CEO, Recomaze
delian@recomaze.ai
Full data set · complete methodology · additional findings · high-resolution charts · interviews.
Go deeper, or check your own store.
Read the complete study
All seven findings, every category broken down, the head-to-head wins, and the full methodology behind the numbers.
recomaze.ai/state-of-AI-commerce-2026 Free toolCheck your own store against AI
Run the same scan we ran, against your real competitors. See which side of the 60% you're on — free, in two minutes, no account.
recomaze.ai