> ## Documentation Index
> Fetch the complete documentation index at: https://docs.fixaeo.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Metrics overview
> The four numbers FixAEO tracks, and what each one answers.
Four metrics, four different questions. Here's how they fit together.
FixAEO measures your brand in AI answers with four metrics: Visibility Score, share of voice, position, and sentiment. They share one raw material — a captured answer per prompt, per engine, per region, per day — but each divides it up differently.
## What does each metric answer?
| Metric | The question it answers | Denominator | Scale |
| --------------------------------------------- | ------------------------------- | ------------------------ | ------------------------------ |
| [Visibility Score](/metrics/visibility-score) | Am I named at all? | Answers captured | 0–100% |
| [Share of voice](/metrics/share-of-voice) | Am I named more than my rivals? | All brand mentions | 0–100% |
| [Position](/metrics/position) | Am I named early or late? | Answers that mention you | 1 and up, lower is better |
| [Sentiment](/metrics/sentiment) | Is the mention any good? | Mentions of your brand | Positive, neutral, or negative |
## Which one should you watch?
Start with [Visibility Score](/metrics/visibility-score). It's the headline number on your [dashboard](/dashboard), and it has to move before the others mean much. You can't rank well in an answer you're absent from.
Once you're named consistently, [share of voice](/metrics/share-of-voice) becomes the more useful number. It's the only one of the four that accounts for what your competitors are doing.
Treat [position](/metrics/position) and [sentiment](/metrics/sentiment) as quality checks on the first two. High visibility with a weak position means you're the afterthought in every list. High visibility with negative sentiment is worse than not being named.
These four can disagree, and that's the point. If your Visibility Score climbs while share of voice drops, the engines are naming more brands overall — you included, but so is everyone else.
## Where do the numbers come from?
Every metric comes from captured answers, not from estimates or modeling. FixAEO asks each AI engine your tracked prompts, reads the answer a real user would see, and counts what's in it.
That makes your prompt list the denominator behind all four numbers. Get it right first — see [Setting up your prompts](/setting-up-prompts).
Because every engine is asked separately, some numbers can also be read one engine at a time. [Visibility Score](/metrics/visibility-score) splits per engine on the Model page, and [share of voice](/metrics/share-of-voice) splits behind the engine filter on [Industry benchmarks](/industry):
Each metric page defines its number and tells you how to read it, including the cases that show a blank instead of a zero. [Understanding AI answers](/understanding-answers) covers the sampling behind all four.
Where each metric appears, and which panel to read first.
The raw captured answers behind every number.
Share of voice, ranking, and the competitive quadrant.
Turn a weak metric into work you can ship.