Before you start
- A live HTTPS site with honest, people-first content.
- Basic familiarity with sitemaps and Search Console (helpful but not required).
Understand ads vs organic citations
OpenAI shows Sponsored cards below some ChatGPT responses — they are labeled and separate from the model answer. Paying for ads does not insert your brand into the organic reply text.
GEO targets citations: when a user asks “how do I set up WordPress hosting in the EU?”, models pull from indexed, crawlable pages with clear steps and verifiable facts.
Official: https://help.openai.com/en/articles/20001047-ads-in-chatgpt
Publish crawlable HTML with one canonical URL
LLM crawlers (GPTBot, ClaudeBot, PerplexityBot) need server-rendered text — not empty SPAs or content locked behind login.
Pick one canonical hostname (https://example.com) and redirect www/alternate hosts. Duplicate URLs dilute what models learn about your site.
Allow useful bots in robots.txt; block only what you must for abuse, not legitimate AI crawlers you want to cite you.
Structure pages as HowTo with clear steps
Use numbered steps, descriptive H2/H3 headings, and short paragraphs — the same patterns that help humans help models extract facts.
Add HowTo schema when steps on the page match the schema exactly; misleading structured data hurts trust.
Link related guides internally so crawlers discover your topic cluster.
Ship llms.txt with your best URLs
llms.txt is a plain-text index of pages you want models to prioritize — hub page, key product pages, and 5–10 flagship guides.
Host it at /llms.txt and optionally /.well-known/llms.txt. Keep it updated when you publish new cornerstone content.
Example hub: https://neytom.com/llms.txt
Measure mentions, not vanity traffic
Track branded search, referral headers from AI products where visible, and manual spot-checks in ChatGPT/Perplexity for queries you care about.
Harvest traffic (single-hit scrapers from sitemap) is not GEO success — focus on durable pages that answer real questions.