By Carson Cobb. Updated November 2026.
A lot of what gets written about AI search is opinion. This page collects the numbers: what the big platforms and analytics companies have published about AI search and AI shopping in 2026, and what my own tests of ChatGPT, Gemini and Google AI Mode found. Every figure has a date and a source. I’ll keep it updated.
How many people use AI search
- Google AI Mode passed 1 billion monthly users, and the Gemini app reached 950 million monthly users, Alphabet said on its Q2 2026 earnings call in July 2026. (MarketBeat)
- ChatGPT had about 900 million weekly users and about 50 million paying subscribers in early 2026, according to OpenAI’s head of ChatGPT. (ROIC)
How shoppers use AI
From a 2026 survey of 1,009 US consumers by Exploding Topics, a Semrush company (Search Engine Land):
- 77.6% had used AI to shop in the past six months. 43% use it weekly or more.
- 68.5% use it for product research, the most common use, and 55% to find the best price or deals.
- ChatGPT was used by 77.6% of AI shoppers and Google Gemini by 58.2%.
- 68.6% said AI had led them to buy something they otherwise wouldn’t have.
- Shoppers still want to buy themselves. “Skeptical” (41%) and “suspicious” (33%) led reactions to AI tools that complete orders. The median amount people would let AI spend on its own was $50.
From an Adobe survey of 5,000 US online shoppers in 2026 (Digital Commerce 360):
- 77% of AI users said AI made them more confident in their purchases.
- 69% said they were less likely to return items.
AI traffic and conversion
From Adobe Analytics, based on more than 1 trillion visits to US retail sites (Adobe, October 29, 2026):
- AI traffic to US retail sites grew 71% year over year in September 2026.
- AI traffic converted 90% better than non AI traffic in September 2026, the highest gap Adobe has recorded. In August 2025 it converted 9% worse.
- Revenue per visit from AI traffic was 83% higher.
- AI visitors spent 73% more time on site, viewed 25% more pages and were 53% more likely to add products to the cart.
- The average US retail homepage scored 63% on Adobe’s citation readability score, a measure of how much page content AI can understand and cite.
The 2026 holiday season
From Adobe’s 2026 holiday forecast (Digital Commerce 360, September 29, 2026):
- US online holiday sales forecast at $275.1 billion, up 6.7%.
- AI traffic forecast to rise 130% year over year over the season, including 159% on Thanksgiving, 95% on Black Friday and 88% on Cyber Monday.
- AI referral traffic in July 2026 was up 1,219% compared with October 2024.
Shopify
- AI referred visitors who landed on a product page converted about 49% more often than organic search visitors in Q1 2026, and orders from AI search were 14% larger. (Shopify, May 2026)
- More than half of AI referred sessions began on a product page, compared with about 20% for organic search. (Shopify, May 2026)
- Organic search still sent more sessions to Shopify merchants than all tracked AI platforms combined. (Shopify, May 2026)
- AI referral sessions grew more than 8x and AI referred orders nearly 13x year over year in Q1 2026. (PYMNTS)
- AI driven traffic and orders each roughly tripled year over year in Q2 2026. (Retail TouchPoints)
- AI attributed traffic grew 2.6x and AI attributed orders 3.3x year over year in Q3 2026. (AI Weekly, November 3, 2026)
- Shopify’s Agentic Storefronts publish merchant catalogs to ChatGPT, Perplexity and Microsoft Copilot through Shopify Catalog, announced in December 2025. (PYMNTS)
What AI recommends: my research
From my tests of ChatGPT, Gemini and Google AI Mode in November 2026, using 48 buyer questions across apparel, beauty and skincare, food and beverage, home, pet and supplements:
- AI names few brands. Median brands per answer: ChatGPT 4, Google AI Mode 5, Gemini 6. (ChatGPT vs Gemini vs Google AI Mode)
- The three AIs mostly disagree. Only 23% of brands named came from all three, and 53% from just one. They picked the same top brand in 15 of 45 questions. (ChatGPT vs Gemini vs Google AI Mode)
- The top pick is stable. The rest isn’t. On repeat runs of the same question, ChatGPT’s first pick came back 98% of the time, but brands in fifth place or lower only 56%. (How often AI recommendations change)
- Wording changes the answer more than randomness does. Rewriting short questions the way shoppers really ask kept the same top pick in only 9 of 24 questions on ChatGPT and 8 of 24 on AI Mode. (How question wording changes AI recommendations)
- Google AI Mode links to brand websites more than anything else. 39% of its links went to brand sites and 6% to retailers. For specific questions, brand sites got 49%. (Where AI sends shoppers)
- Reddit appeared in 19 of 48 AI Mode answers and YouTube in 18. (Where AI sends shoppers)
- Ratings act as a minimum bar. 96% of rated products on AI Mode’s product cards were 4.0 stars or higher, but the first product shown was the highest rated in only 5 of 21 answers. (Do star ratings decide who AI recommends?)
What AI gets wrong: my research
- Prices are mostly right. ChatGPT and Gemini gave the right price for 22 of 24 Shopify products, AI Mode for 19. (What AI gets wrong about your products)
- Ratings often aren’t. All three AIs got a product’s rating and review count right for only 7 of 22 products. (What AI gets wrong about your products)
Shopify product data: my research
From 94 US Shopify stores checked in November 2026:
- The typical store scored 70 out of 100 on AI readiness. (Shopify AI Readiness Index)
- 29% of stores publish duplicate product data, usually from a review app. (Duplicate product data)
- 30% of sampled products had no barcode on any variant, rising to 63% in food and beverage. (Is your Shopify product data ready for AI shopping?)
- Only 43% of sampled products passed four basic product data checks. (Is your Shopify product data ready for AI shopping?)
- 23% of product pages explain how to use the product, and 16% compare it to anything. (Do your product pages answer what shoppers ask AI?)
What the numbers add up to
AI is a small share of traffic and the fastest growing one, and the shoppers it sends arrive closer to buying. They land on product pages more than any other page, and they’ve already been told what to expect by an AI that names only a handful of brands, differs from tool to tool, and often gets ratings wrong.
For a Shopify brand, that means three jobs: get into the short list on each AI, make sure AI describes your products correctly, and make the product page confirm what the shopper was told. That’s the work behind my AI Search and Conversion Audit. For a quick first look, run the free AI Visibility Check.
About these numbers
Figures from other organizations are quoted as published, with the source and date linked. Some are reported by news outlets summarizing company announcements and earnings calls, and the original wording may differ slightly. My own figures come from the reports linked beside them, each with its full method.