> ## Documentation Index
> Fetch the complete documentation index at: https://tryprefer.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# How Prefer Works: The AI Search Measurement Loop

> Prefer runs your tracked prompts across every major answer engine on a schedule, scores each answer, and turns the gaps into ranked, evidence-backed actions.

Prefer measures your brand's presence in AI answer engines by running a fixed set of prompts against every enabled engine, on a schedule, and scoring what comes back. This page explains that loop end to end so you can reason about what a metric change means and where each number originates.

## The four stages

<Steps>
  <Step title="Prompts">
    Your project maintains a list of **tracked prompts**: real buyer questions grouped by **topic** and, optionally, by **persona** and **region**. Prompts are generated during onboarding, added from the curated library, or drafted with the AI assistant in the Prompts console.
  </Step>

  <Step title="Runs">
    On a schedule, Prefer executes a **tracking run**: one full pass that sends every enabled prompt to every enabled engine. You can also trigger a run manually from the top bar with **Run tracking now**.
  </Step>

  <Step title="Extraction">
    For each answer, Prefer stores the full text, extracts every brand mentioned, records every cited URL, and computes sentiment. Your brand, your competitors, the cited sources, and the position you appear in are all normalized so they can be aggregated.
  </Step>

  <Step title="Scoring and actions">
    Prefer rolls the extracted signals into **Visibility Score**, **Share of Voice**, **Citation Share**, **Avg Position**, and **Sentiment**, per engine and blended. Gaps between your performance and the leader's feed the Action Center as ranked, evidence-backed work items.
  </Step>
</Steps>

## Runs vs the free audit

<CardGroup cols={2}>
  <Card title="Free audit" icon="magnifying-glass" href="/docs/get-started/free-audit">
    A single one-time pass. About 30 prompts across 3 engines. Produces a report and a starter action plan. Always free.
  </Card>

  <Card title="Ongoing tracking" icon="gauge" href="/docs/monitor/overview">
    Repeats on a schedule on every paid plan, across all five engines and your plan's prompt count. Feeds trends, deltas, alerts, and the Action Center.
  </Card>
</CardGroup>

## How deltas work

Every metric card shows a delta labelled "vs previous run", not "vs yesterday". A run is the unit of time in Prefer, because engines can shift between hours and daily aggregation would hide those movements. If you run tracking twice in one day, you will see two deltas.

<Tip>
  Use the Filter Bar on any Monitor page to narrow scope to a single engine, topic, persona, region, or tag. Every metric on the page recomputes to that scope.
</Tip>

## Where each number comes from

| Metric | Source |
| - | - |
| Visibility Score | Share of tracked answers in scope that name your brand |
| Share of Voice | Your mentions divided by all brand mentions in scope |
| Avg Position | Mean position among named brands, inverted so 1 is best |
| Citation Share | Citations to your domain divided by all citations |
| Sentiment mix | Polarity assigned to each mention in scope |

## Next

<CardGroup cols={2}>
  <Card title="Metrics" icon="chart-line" href="/docs/concepts/metrics">
    Full definitions and how each metric is computed.
  </Card>

  <Card title="Engines" icon="robot" href="/docs/concepts/engines">
    Which engines Prefer measures and what your plan includes.
  </Card>
</CardGroup>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.