How citeme.eu works
The three surfaces we measure, how prompts and search queries turn into metrics, and how often data is collected.
Three surfaces, measured separately
| Surface | The question it answers | Driven by | Where in the app |
|---|---|---|---|
| AI visibility | Does an AI assistant's answer name you? | Prompts | AI Visibility, Brand Profile |
| Search visibility | Does a search engine list your page — or someone else's page that names you? | Search queries | Search Visibility |
| Site audit | Can AI systems reach, read and cite your website? | Your website | Site audit |
The three move independently, and fixing one does not fix the others. That is why citeme.eu tracks them side by side, over the same period, instead of in three separate tools. See AI visibility vs search visibility for why the difference matters.
From a question to a number
- You choose what to track. Prompts are the questions your buyers ask AI assistants — "What is the best HR software for a 50-person team?". Search queries are what they type into a search engine. Both are grouped into topics. When you sign up we suggest a starting set based on your website, and you can change all of it.
- We ask on a schedule. Every active prompt is sent to each AI assistant in your plan, with the market and language you set for it, wherever the assistant supports them. Every active search query is run against the search engines the same way: Google and Bing get both, the AI search engines only the market. Google usually derives the language from the region, and you can override it. ChatGPT cannot, and follows the language of the text. See Prompts and topics and Search queries.
- We read every answer. For each AI answer we record which brands it names and in what order, how it describes your brand, which web pages it cites, and which web searches the chatbot ran while writing it. For each search results page we record where your site ranks, which third-party pages name you, and whether Google showed an AI Overview above the links.
- We turn readings into metrics — visibility, share of voice, position, sentiment and more — following the rules on How we calculate.
- We suggest next steps. Recommended actions appear on Overview, AI Visibility, Search Visibility and Brand Profile. Each one includes the reasoning and concrete steps. Overview picks at most the three most important, including findings from the site audit.
Every answer is stored. Behind any number you can open the answers it was computed from and read exactly what the assistant said and which sources it used.
What we track
AI assistants — ChatGPT, Google AI Mode, Gemini and Perplexity. Your plan decides which of them are asked. Results are combined across them by default, and every page can be filtered to a single assistant.
Search engines — traditional search (Google, Bing) and agentic search built for AI systems (Tavily, Exa, Perplexity Search). For Google we also capture the AI Overview above the links.
Websites — a technical audit of your site and of every competitor you track, up to 40 pages per site.
How fresh the data is
- Prompts and search queries run on your plan's schedule: three, five or seven days a week. A new prompt can be run straight away, and a newly tracked search query starts running immediately.
- The site audit runs every week for your site and your competitors' sites. You can also start one at any time.
- Your brand details — what you offer and to whom — are refreshed from your website once a month, so suggestions for new prompts and search queries keep up with your business.
Every page shows the past three weeks by default. You can choose any other date range: a shorter one reacts faster but rests on fewer answers, a longer one is steadier.
When there is too little data
A percentage built on a handful of answers is noise. When a page doesn't have enough data for a reliable conclusion, it says it is still collecting instead of showing a verdict — for example, AI Visibility waits for at least 10 answers and Search Visibility for at least 20 search runs. The exact thresholds are listed on How we calculate.