GEO emerged around 2023 as a concept: a significant share of search is moving from "keyword → link" to "question → LLM answer," so brands/content must optimize to be cited by LLMs.
Traditional SEO vs GEO
| | SEO | GEO | |---|---|---| | Target | Search engines (Google, Naver, Bing) | LLMs (ChatGPT, Claude, Perplexity, Copilot) | | Core | Links, keywords, backlinks | Structured data, fact density, citability | | Metric | Search rank | Citation frequency inside LLM answers |
Core techniques
- Publish **llms.txt**: site summary for LLM crawlers - **JSON-LD structured data**: DefinedTerm, SoftwareApplication, FAQPage, etc. - **AI bot allowlists in robots.txt**: explicitly allow GPTBot, ClaudeBot, PerplexityBot - **Fact-dense content**: concrete numbers and definitions over vague marketing copy - **Locale-independent content**: write native copy per locale, not translations - **hreflang x-default**: for multilingual sites - **Register with Bing Webmaster**: ChatGPT and Copilot use Bing's index - **IndexNow submission**: notify Bing, Yandex, Seznam instantly
Korea/Japan specifics
- **Korea**: ChatGPT and Perplexity learn from English data, so native Korean content is essential. Run alongside Naver search - **Japan**: Copilot and Claude usage rising. Japanese honorifics + fact-first writing boosts citation probability
Measurement tools
- Query ChatGPT/Claude/Perplexity and check for citations - Bing Webmaster crawl log → GPTBot / OAI-SearchBot hits - Google Search Console → baseline SEO - Audit structured docs like crewtool.app/robots.txt
The Crewtool project itself is a live GEO experiment.