Method: how to build a site AI assistants cite
ChatGPT, Alice, Gemini and Perplexity answer questions based on the pages they find. A page that is easier for a machine to read and verify has a better chance of being cited.
Six rules
Text served in HTML
Pages arrive with their text already rendered. A crawler does not need to run JavaScript to understand the site.
schema.org markup
Our structured data reflects the company, products and services described on the page. Standard Google Search requirements apply to AI Overviews and AI Mode.
Questions and answers
The FAQ block uses the words buyers use. A short, precise answer is the easiest thing to cite with a link.
An llms.txt file
A short brief for language models: what is sold, where, the key sections and links.
Open robots.txt and sitemap
A sitemap helps crawlers discover pages. OAI-SearchBot supports ChatGPT search. GPTBot is used for model training and its access is configured separately.
One page, one question
A separate page for each product, application, country and common question, so it matches the query closely.
What we do not promise
- A place in a specific assistant's answer: the algorithms are closed and change
- Results without content: query pages are written on a schedule
- The same timeline for every niche
How we measure
AI-assistant visits show up in Yandex Metrica's Sites report: chatgpt.com, alice.yandex.ru, gemini.google.com, perplexity.ai. We count visits that reach selected contact goals. We verify received enquiries separately. Google impressions and clicks come from Search Console.
Reports count referrals with a known source. If it is lost, the chat cannot be identified from the visit. The llms.txt file lists resources. We verify citations separately by date, question and link.
Google Search Central · OpenAI
Research on our data →We will build your site by these rules
Choose how to pay. Every option includes the same site and timing.