What is the MentionX methodology?

MentionX asks AI engines real buyer questions, stores every answer, extracts mentions, rankings and citations, then calculates metrics with versioned formulas in code. AI models extract what an answer says; they never calculate a score.

Does MentionX use AI to score brands?

No. A language model reads each stored answer and fills a strict schema: who is mentioned, in what position, with which citations. Every metric is then calculated by fixed formulas in MentionX’s own code, so the same evidence always gives the same number.

How is the MentionX Visibility Score calculated?

Visibility Score = 0.40 × Mention Rate + 0.35 × normalised position + 0.25 × citation share (scorer-v0.2.0). The weights are constants in code and are published here whenever they change.

How does MentionX choose buyer prompts?

You enter your URL and market. MentionX finds your category and competitors, then writes prompts across six intents: 30% category discovery, 20% use cases, 15% buying criteria, 15% competitive, 10% alternatives and 10% brand. You can edit, import and add prompts at any time.

Does MentionX scrape ChatGPT?

No. MentionX uses official APIs wherever an engine offers one, in fresh sessions with no login, history or memory. Google AI Overviews and AI Mode have no official API, so they come through a licensed search-results API. Microsoft Copilot answers come from Microsoft’s Azure API with Bing grounding.

Why does MentionX scan weekly instead of daily?

AI answers vary from one run to the next, so daily numbers mostly measure noise. Four full scans a month on the same prompts show real movement and keep results comparable.

Can I check how a number was calculated?

Yes. Every metric opens to the stored answers behind it, with the engine, prompt, date and citations. Each scan also records its prompt set and the version of every step, so past results can be recomputed exactly.