We asked ChatGPT, Gemini, Perplexity and Naver AI Tab 136 buying questions in Korea and the US. 82% of the 479 answers named a brand. Here is who they named.
What the AI Index measures
The AI Index is a set of category monitors that querying.ai runs on itself. Each category is a fixed list of buying questions, written the way people ask an assistant and chosen from what people in that market search for. Every week each engine answers every question from inside the market's country, and each answer is scored for the brands it names.
This report is the snapshot of the run on September 30, 2026: 14 categories, seven in South Korea and seven in the United States, 136 questions and 479 answers. Korean questions went to ChatGPT, Gemini, Perplexity and Naver AI Tab, and US questions to ChatGPT, Gemini and Perplexity. Each engine answered each question once.
Two numbers describe the answers. A brand's rate is the share of answers that name it. The named rate is the share of answers that name any brand at all.
How often AI answers name a brand
Most answers to buying questions recommend brands by name. How often depends on the engine and the market:
| Market | Engine | Answers | Named a brand | Rate | 95% interval |
|---|---|---|---|---|---|
| South Korea | ChatGPT | 71 | 43 | 61% | 49-71% |
| South Korea | Gemini | 71 | 63 | 89% | 79-94% |
| South Korea | Perplexity | 71 | 54 | 76% | 65-84% |
| South Korea | Naver AI Tab | 71 | 68 | 96% | 88-99% |
| United States | ChatGPT | 65 | 56 | 86% | 76-93% |
| United States | Gemini | 65 | 57 | 88% | 78-94% |
| United States | Perplexity | 65 | 53 | 82% | 70-89% |
Korean ChatGPT stands apart. Its interval (49-71%) sits entirely below Naver AI Tab's (88-99%) and Gemini's (79-94%), so the gap is larger than sampling error. In Korean supplements, ChatGPT named a brand in 4 of 10 answers and Naver AI Tab in 10 of 10. In Korean skincare it was 5 of 12 against 10 of 12.
Who leads each category
The three brands each category's answers named most, with the share of answers that named them:
| Market | Category | Answers | First | Second | Third |
|---|---|---|---|---|---|
| South Korea | Credit cards | 40 | Shinhan Card 78% | Hyundai Card 35% | KB Kookmin Card 28% |
| South Korea | Flights | 36 | Skyscanner 69% | Naver Flights 47% | Trip.com 42% |
| South Korea | Travel stays | 40 | Yanolja 65% | Yeogi Eottae 63% | Agoda 58% |
| South Korea | Laptops | 44 | ASUS 61% | Lenovo 61% | LG gram 45% |
| South Korea | Sneakers | 36 | ASICS 53% | Nike 50% | New Balance 50% |
| South Korea | Supplements | 40 | Centrum 30% | Alive 20% | Korea Eundan 18% |
| South Korea | Skincare | 48 | Round Lab 19% | Dr.G 17% | La Roche-Posay 13% |
| United States | Running shoes | 21 | On 100% | HOKA 71% | Nike 67% |
| United States | Laptops | 30 | Lenovo 93% | ASUS 63% | HP 63% |
| United States | Flights | 27 | Google Flights 89% | Skyscanner 81% | Kayak 70% |
| United States | Credit cards | 30 | Capital One 73% | Chase 70% | American Express 53% |
| United States | Skincare | 27 | CeraVe 48% | La Roche-Posay 33% | The Ordinary 30% |
| United States | Travel and hotels | 30 | Expedia 43% | Booking.com 40% | Google Hotels 37% |
| United States | Supplements | 30 | Ritual 17% | Thorne 17% | Culturelle 10% |
Three patterns stand out.
- Clear defaults. In the US, On appeared in all 21 running shoe answers, Lenovo in 28 of 30 laptop answers and Google Flights in 24 of 27 flight answers. When one brand is in nearly every answer, the open competition is for the second and third places.
- Local platforms in Korea. Korean travel answers named Yanolja (65%) and Yeogi Eottae (63%) ahead of Agoda (58%), and Korean flight answers named Naver Flights in 47%. The same category in the US named Expedia, Booking.com and Google Hotels.
- Spread-out categories. Supplements had no brand above 30% in either market, and US supplements had the lowest named rate of all 14 categories: 14 of 30 answers named a brand.
Engines pick different leaders
Inside one category, each engine has its own favorite. In Korean skincare the brand named most often was Round Lab on ChatGPT, Innisfree on Gemini, CeraVe on Perplexity and Dr.G on Naver AI Tab. In Korean sneakers it was Nike on ChatGPT and Naver AI Tab, ASICS on Gemini and New Balance on Perplexity. In US travel, ChatGPT named Expedia most and Gemini and Perplexity named Booking.com most.
Each engine's leader rests on 9 to 12 answers, which makes it a direction to confirm with your own monitor. The direction is consistent, though: each engine is its own channel, so measure every engine your buyers use, in every market you sell in.
How to read these numbers
Every rate here comes from one run, and AI answers vary from run to run. With 48 answers, a 50% rate carries a 95% margin of about 14 points; with 21 answers, about 21 points. The engine totals (65 to 71 answers each) are firmer, and the overall 82% (479 answers) holds within about 3 points.
Read a category's ranking as a shortlist and follow it over several weeks before you act on a single gap. The sample size guide shows how many answers each kind of decision needs.
Build the same report for your category
The AI Index runs on the same monitors every querying.ai account can create. A category monitor takes a category, its sub-fields and up to 25 brands to rank, and reports each brand's share of answers by engine, by sub-field and over time. A brand monitor follows your own brand and its competitors, with mention and citation rates and alerts.
The board's data is public JSON, refreshed with every weekly run:
curl https://api.querying.ai/v1/ai-indexFrequently asked questions
What is the AI Index?
A public board of category monitors that querying.ai runs every week in South Korea and the United States. It ranks the brands ChatGPT, Gemini, Perplexity and Naver AI Tab name when people ask buying questions in each category.
How are the questions chosen?
Each category's questions come from search demand in its market and are written the way people ask an AI assistant. The same questions run every week, so changes over time come from the engines.
How often is the AI Index updated?
Each category runs weekly. The AI Index page shows the last 30 days and links every category to its full report, with the answers behind each number.
Can I measure my own brand the same way?
Yes. Create a monitor with your prompts, engines, country and schedule. Each run scores every answer for your brand, your domains and your competitors.
