Mention Meter
AI search visibility

Do AI assistants recommend you?

Mention Meter asks the AI assistants your buyers use the questions they actually ask, many times over, then measures who gets recommended, in what order, and why.

Prompt × platformExample
ChatGPTClaudeGeminiPerplexityAI Mode

Illustrative data. Each cell shows how many runs of a prompt named the company.

How it works

A repeatable measurement, so a baseline can be compared with a re-measurement after work to improve visibility.

  1. Set up

    Add the client, its other names and websites, the facts about it, and a set of buyer-intent prompts.

  2. Collect

    Every prompt runs several times on each platform, each run a fresh session with web search on.

  3. Analyse

    Each answer is read to find the companies named, their order, what they're credited for and the sources cited.

  4. Review

    Every score can be checked against the answer and corrected, and duplicate companies merged.

  5. Report

    An interactive report is published at a private link for the client.

Platforms

Four platforms are queried automatically through their APIs. Google AI Mode is collected by hand from a guided queue.

ChatGPTAPI ClaudeAPI GeminiAPI PerplexityAPI Google AI ModeManual

Features

Everything needed to run a measurement from prompts to published report.

Prompt sets

Paste prompts in bulk with categories, and mark branded prompts that name the client for fact-checking.

Automatic collection

Runs in the background with per-platform rate limits, retries, and resume or retry for anything that fails.

Manual entry queue

One prompt at a time with copy buttons and a direct AI Mode link, so hand-collected answers are quick and consistent.

Cost tracking

Tokens, web searches and cost recorded for every response, with a running total and estimate during collection.

AI analysis

Finds every company named, its position in the recommendation list and what it's credited for, with a name check as a safety net.

Fact checks

Answers to branded prompts are checked against the client's facts: correct, partly right, wrong or not mentioned.

Review screen

The answer with names highlighted beside an editable scores table. Corrections survive re-analysis.

Competitor list

Every company named across the report, with duplicates merged so each one is counted once.

Private report link

Summary, prompt grid, competitors, sources, fact checks and every raw response, behind a link that can be revoked.

Scoring

Each answer gets a score for the client, which adds up to appearance rate, average score, times named first and share of voice.

0

Not mentioned

1

Mentioned in passing, or its site is cited

2

Named in a recommendation list

3

First or top recommendation