# Metric definitions

> Every score and label in a Cited report, in one place.

Source: https://youcited.com/docs/geo-concepts/metric-definitions
Section: GEO Concepts
Last updated: 2026-10-29
All docs: https://youcited.com/docs/index.md

## Metric Definitions

Every term you'll see in a Cited report, in plain words. For how they fit together, see [Understanding your scores](/docs/getting-started/understanding-your-scores).

## Scores

### Model score (0-100)

One score per AI model: Cited counts for 50%, Visibility for 30% and Share of Voice for 20%.

### Engine score (0-100)

The average of the model scores inside one engine (ChatGPT, Gemini, Claude or Google AI). Where an engine has as-seen results, only those count.

### Overall score (0-100)

The average of the engine scores, with every engine weighted the same unless you change the weights.

### Site-check score (0-100)

How well your site gives AI what it needs, from the [site check](/docs/feature-guides/site-check). Separate from the AI scores.

### Bands

Not showing up (0-19), Weak (20-39), Fair (40-59), Strong (60-79), Leading (80-100).

## The three parts

### Cited

The share of answered questions where the model links to your site as a source.

### Visibility

The share of answered questions where you show up at all: linked, named, or among the pages the model read. Always at least as high as Cited.

### Share of Voice

Your site's share of all the sites the model cited across your questions. If the model cited 40 sources in total and 4 were yours, your Share of Voice is 10%.

## Per-question labels

### Cited in this answer

The answer links to your site.

### Seen, not cited

The answer names you or read your page, but doesn't link to you.

### Cited instead

The sites the answer cited when it didn't cite you, or alongside you.

### Runs

In a paid audit each question is asked three times per model. "2 of 3" means two of the three answers cited you. The row shows the majority result.

### How much of the answer draws on your site

A rough measure of how much of an answer's text comes from your pages: Low, Some or High. A second version counts early mentions more, since readers see those first.

## Methods

### API

The model asked through its developer API with web search on.

### As seen

What a real person sees in the ChatGPT or Gemini app, collected through a browser. Where we have it, it counts instead of the API result.

## Recommendations and fixes

### Proven

Cited's citation research shows this kind of change wins AI citations. A strength word says how strong the evidence is.

### Best practice

A widely accepted GEO practice we can't yet back with our own data.

### Fix stages

Ready (files and guide made), Done (you added it), Verified (Cited found it on your live site), Working (a later audit shows it helped).

## Money

### Credit

1 credit = $1. See [Credits and pricing](/docs/getting-started/credits-and-pricing).