by FlowHunt
English EN
English English Deutsch Deutsch Français Français Español Español Italiano Italiano Slovenčina Slovenčina Čeština Čeština Dansk Dansk 日本語 日本語 Nederlands Nederlands Norsk Norsk Polski Polski Português Português Română Română Svenska Svenska
 Get Started
Documentation
MCP Glossary Open app

Core metrics

Metrics overview

The core AmICited metrics in one place: what each one measures, how it is calculated, and how to read them together to diagnose what is holding your AI visibility back.

Open in ChatGPT Open in Claude

AmICited turns every stored AI answer into a small set of metrics. Each answers a different question, and most confusion comes from reading one metric when you needed another. This page gives you the short version of each, then shows how to combine them into a diagnosis.

The two building blocks#

Every metric is built from one of two records taken from each answer:

  • Brand mentions. The brands, companies and products the answer names in its text, linked or not, in order of first appearance. Each brand counts once per answer.
  • Citations. The URLs the engine listed as sources, each with its position in the source list.

A brand can be named without its site being cited, and a page can be cited without the brand being named. The two move for different reasons and are fixed by different work. If you read one page in this section, make it mentions vs. citations.

The core metrics#

MetricQuestion it answersBuilt fromFormula
Visibility scoreDo we appear at all?Mentionsprompts where you were named ÷ tracked prompts × 100
Share of voiceHow much of the conversation is ours?Mentionsyour mentions ÷ all brand mentions × 100
Average rankWhen we appear, how early?Mentionsmean position of your mentions (1 = named first)
Share of citationAre our pages used as sources?Citationscitations of your domain ÷ all citations × 100
Not rankedWhich questions do we lose completely?Mentionstracked prompts − prompts where you were named

Two counts sit beside them on the dashboard:

  • Total citations. Despite the label, this counts brand mentions: the number of answers that named your brand in the period. It is the numerator of share of voice.
  • Tracked prompts. How many prompts match the current filters. It is the denominator of visibility.

All percentages are rounded to one decimal place. Every metric follows the dashboard filters (date range, engine, country, tag) and is compared with the previous period of the same length. Changes in percentages are shown in percentage points (pts).

Reading them together#

No single number tells you what to do. The combination does:

PatternWhat it usually meansWhere to act
Low visibility, good average rankWhere you appear you lead, but you are missing from most prompts. A coverage problem.Work the not-ranked prompts; check prompt coverage.
High visibility, low share of voiceYou appear in most answers, but competitors are named far more often, usually as one name in a long list.Compare with competitors; get onto the pages engines cite.
High visibility, poor average rankYou are named, but late: a footnote after the recommendations.Strengthen comparison and “best X” content; see average rank.
High share of voice, low share of citationEngines talk about you from third-party sources, not your own pages.Find the cited pages in Sources; give engines a page of yours worth citing.
High share of citation, low share of voiceYour pages feed answers that recommend someone else.Make your brand explicit on the cited pages; check prompt detail.
Metrics swing wildly day to dayFew prompts, few runs, or one noisy engine.Filter by engine; widen the date range; add prompts.

Worked example#

Over 30 days you track 40 prompts. Your brand is named in answers to 14 of them, 62 times in total, out of 480 brand mentions across all answers. Your average position is 2.8, and your domain has 35 of the 700 citations.

  • Visibility: 14 ÷ 40 × 100 = 35.0%
  • Share of voice: 62 ÷ 480 × 100 = 12.9%
  • Average rank: #2.8
  • Share of citation: 35 ÷ 700 × 100 = 5.0%
  • Not ranked: 40 − 14 = 26 prompts

Reading: when you are named you are near the top (rank 2.8), but you are absent from 26 prompts and your own pages supply only 5% of sources. The priority is coverage, not reputation: start with the not-ranked prompts and the sources engines use for them.

What AmICited does not measure#

  • Sentiment. AmICited records whether and where you are named, not whether the answer is positive about you. Read the answer text on the prompt page to judge tone.
  • Clicks from AI answers. Mentions and citations are measured in the answers themselves. Traffic is a separate question, answered by connecting Google Analytics 4 and Search Console.

Changing the prompt set changes every metric

Adding or removing prompts moves visibility and share of voice even when nothing real changed. When you compare periods, keep the prompt set stable, or filter to a tag that was stable over both periods.

Mentions vs. citations The two records behind every metric. Visibility score The headline coverage metric. Reading the dashboard Where each metric appears, card by card. Why did my score drop? How to trace a drop to its cause.