Introduction
What citat.ai measures, and what the numbers mean.
citat.ai is a generative engine optimization (GEO) platform. It measures how your brand shows up when people ask AI for answers — how often you are mentioned, how often you are cited as a source, and how that compares against your competitors.
Answer surfaces
An answer surface is anywhere a person gets an AI-written answer. citat.ai covers four kinds:
- LLM clients — captured through the web UI, which also exposes the sub-queries a model fans out to (query fanout).
- LLM APIs — the model answering from memory, with no browsing and no fanout. Reading the two side by side separates what a model already knows about you from what it looks up.
- Search AI summaries — Google AI Overviews and AI Mode.
- Social media — AI summaries and search autosuggest.
Entity normalization
The same brand gets written many ways: Chinese and English, full name and abbreviation, aliases, sub-brands. citat.ai normalizes mentions against an identity set that you define yourself, and only then counts. You declare what counts as "you" — we do not guess.
Honest boundaries
Query fanout is only observable on chat clients. It is not available on Google's search surfaces — so we do not report it there. We would rather name the gap than imply coverage we do not have.
Where to go next
This documentation is still being written. The quickstart, the concept reference, the
measurement methodology, and the developer reference (Public API and MCP) are tracked in
VIR-34 through VIR-38.