# How AgentLocal measures local AI visibility

> AgentLocal combines business details linked to public sources with repeatable test searches. It reports what AI assistants return and which public links appear, without claiming access to private prompts or hidden ranking systems.

- Canonical HTML: https://agentlocal.org/research/methodology
- JSON: https://agentlocal.org/api/v1/research/methodology
- Last reviewed: 2026-08-14

## What we collect

- Business details supported by a public source, with checked dates.
- Controlled, non-private test searches representing local-service needs.
- Whether an answer covered the requested details, named an identifiable business, and included a public source link.
- When business details and local guides were last checked.

## Interpretation

- Every percentage shows the count behind it.
- A citation is a public URL present in a measured answer, not proof of hidden model reliance.
- Unmeasured markets are labelled unmeasured, not scored as zero.
- Commercial plans do not change organic ordering or measurement.
- Private prompts and raw IP addresses are not published.

## Publication gate

A local guide needs enough measured interest, multiple matching businesses, recently checked details, enough information to answer the request and at least one current contact or booking option. Pages that no longer meet those standards are removed from public indexing.