What are the minimum requirements to be cited by ChatGPT?

Five baseline requirements. (1) GPTBot allowed in robots.txt. (2) Public access (no login required). (3) Machine-readable Organization / WebSite JSON-LD with legal name, address, URL. (4) Topic-relevant body content (200+ chars). (5) Some external mention from third-party domains. lookupai meets all five on lookupai.jp itself.

Are ChatGPT search and ChatGPT (no web) different?

Yes. ChatGPT search uses OAI-SearchBot for live retrieval — recency and crawl permission matter most. Plain ChatGPT relies on training data — long-running web exposure and content quality matter most. To target both, allow both User-agent: GPTBot and User-agent: OAI-SearchBot in robots.txt, which lookupai does on its own site.

Does llms.txt help ChatGPT citation?

Not yet directly. Anthropic Claude and Perplexity are showing signs of using llms.txt; ChatGPT support is anticipated. In the meantime it serves as a positive signal of LLM-friendly intent and as a quick-reference summary for human reviewers. lookupai ships /llms.txt and /llms-full.txt as part of its standard playbook.

Which JSON-LD types should I implement?

In priority order: (1) Organization, (2) WebSite, (3) FAQPage, (4) Service / Product, (5) Person. Add TechArticle / Article on docs pages and DefinedTerm for glossary entries. lookupai's template is in src/lib/seo.ts in the public GitLab repo.

What writing style gets cited by ChatGPT?

Four principles. (1) Q&A form: question as heading, 200-400 char conclusion-then-detail paragraph. (2) Concreteness: place names, company names, dates, numbers (Hachioji, May 2026, 29,800 JPY, 4 AIs, 40 queries). (3) Self-contained paragraphs: LLMs cite at chunk level, so each paragraph should land its main point. (4) Outbound authority links: Wikipedia, industry reports, trusted sources signal authority chain.

How long until results appear?

Real-time retrieval (ChatGPT search, Perplexity, Gemini web) reflects within 1-4 weeks. Training-data systems (plain ChatGPT, plain Claude) can take months to over a year. lookupai recommends a two-layer monitoring approach: short-cycle re-measurement after each implementation step for retrieval-based systems, and monthly/quarterly trends for training-based systems.

Are secondary-source mentions (Wikipedia etc.) required?

Effectively yes for mid-size and larger brands. Generative models prioritise high-frequency entities as "real and reliable". Increasing exposure on Wikipedia, Qiita, Reddit, hatena bookmarks and trade press is the highest-leverage long-term tactic. lookupai's Pro tier (98,000 JPY/mo) and Premium implementation package (500,000 JPY+) include secondary-source exposure strategy.

DOCS

How to get your company cited by ChatGPT (May 2026)

Becoming a citation source on ChatGPT falls into three layers. (1) Machine-readable foundations (robots.txt / JSON-LD / llms.txt). (2) Content shape (Q&A, concreteness, proper nouns). (3) Secondary-source exposure (Wikipedia, Qiita, Reddit, trade press). This guide walks through each layer with lookupai's own implementation as reference.

Layer 1: foundations

a. robots.txt — explicit AI-crawler allow

ChatGPT (GPTBot) and ChatGPT search (OAI-SearchBot) are different crawlers. Allow both explicitly.

User-agent: GPTBot
Allow: /

User-agent: OAI-SearchBot
Allow: /

User-agent: ChatGPT-User
Allow: /

User-agent: ClaudeBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: Google-Extended
Allow: /

User-agent: Applebot-Extended
Allow: /

Sitemap: https://example.com/sitemap-index.xml

b. JSON-LD structured data

At minimum, ship Organization and WebSite on every page; FAQPage on Q&A sections; Service / Product / Person on the relevant pages; TechArticle / Article on docs. lookupai's templates live in src/lib/seo.ts.

c. llms.txt

A 2024 proposal by Jeremy Howard / Answer.AI to place a Markdown summary at /llms.txt. See our implementation guide.

Layer 2: content shape

Q&A form: question as heading, 200-400 char paragraph. Connect with FAQPage JSON-LD.
Concreteness: place names (Hachioji), legal names (Jimolab LLC), dates (May 2026), prices (29,800 JPY), counts (4 AIs / 40 queries).
Self-contained paragraphs: LLMs cite at chunk granularity.
Outbound authority links: Wikipedia, trusted reports, signal authority chain.

Layer 3: secondary-source exposure

Models prioritise high-frequency entities as "real". Plan deliberate increases in mentions on Wikipedia, Qiita, Reddit, hatena, trade press. lookupai's Pro tier or Premium implementation package includes this strategy work.

lookupai's own implementation

robots.txt: /robots.txt
llms.txt: /llms.txt / /llms-full.txt
This page itself ships TechArticle + FAQPage JSON-LD (inspect via DevTools).
Sitemap: /sitemap-index.xml (auto-generated by @astrojs/sitemap).

Related docs

FAQ

FAQ on ChatGPT citation

What are the minimum requirements to be cited by ChatGPT?

Five baseline requirements. (1) GPTBot allowed in robots.txt. (2) Public access (no login required). (3) Machine-readable Organization / WebSite JSON-LD with legal name, address, URL. (4) Topic-relevant body content (200+ chars). (5) Some external mention from third-party domains. lookupai meets all five on lookupai.jp itself.
Are ChatGPT search and ChatGPT (no web) different?

Yes. ChatGPT search uses OAI-SearchBot for live retrieval — recency and crawl permission matter most. Plain ChatGPT relies on training data — long-running web exposure and content quality matter most. To target both, allow both User-agent: GPTBot and User-agent: OAI-SearchBot in robots.txt, which lookupai does on its own site.
Does llms.txt help ChatGPT citation?

Not yet directly. Anthropic Claude and Perplexity are showing signs of using llms.txt; ChatGPT support is anticipated. In the meantime it serves as a positive signal of LLM-friendly intent and as a quick-reference summary for human reviewers. lookupai ships /llms.txt and /llms-full.txt as part of its standard playbook.
Which JSON-LD types should I implement?

In priority order: (1) Organization, (2) WebSite, (3) FAQPage, (4) Service / Product, (5) Person. Add TechArticle / Article on docs pages and DefinedTerm for glossary entries. lookupai's template is in src/lib/seo.ts in the public GitLab repo.
What writing style gets cited by ChatGPT?

Four principles. (1) Q&A form: question as heading, 200-400 char conclusion-then-detail paragraph. (2) Concreteness: place names, company names, dates, numbers (Hachioji, May 2026, 29,800 JPY, 4 AIs, 40 queries). (3) Self-contained paragraphs: LLMs cite at chunk level, so each paragraph should land its main point. (4) Outbound authority links: Wikipedia, industry reports, trusted sources signal authority chain.
How long until results appear?

Real-time retrieval (ChatGPT search, Perplexity, Gemini web) reflects within 1-4 weeks. Training-data systems (plain ChatGPT, plain Claude) can take months to over a year. lookupai recommends a two-layer monitoring approach: short-cycle re-measurement after each implementation step for retrieval-based systems, and monthly/quarterly trends for training-based systems.
Are secondary-source mentions (Wikipedia etc.) required?

Effectively yes for mid-size and larger brands. Generative models prioritise high-frequency entities as "real and reliable". Increasing exposure on Wikipedia, Qiita, Reddit, hatena bookmarks and trade press is the highest-leverage long-term tactic. lookupai's Pro tier (98,000 JPY/mo) and Premium implementation package (500,000 JPY+) include secondary-source exposure strategy.

Find out where you stand on AI search — free.

Send your URL and 3-5 keywords. Within two business days we deliver a free report covering ChatGPT, Perplexity and Gemini.

Run free audit See pricing