How Citadex's Visibility module tracks your brand mentions across ChatGPT, Claude, Gemini, Perplexity, Copilot, Grok, DeepSeek, and Meta AI. Understand AI Share of Voice, mention rate, sentiment, context health, and platform trends.

AI Visibility Tracking

The Visibility module is Citadex's real-time monitoring engine. It systematically queries AI platforms with your topic-based prompts and measures how consistently your brand is mentioned, how positively it's framed, whether it's described in the right context, and how you compare to competitors.

For per-prompt monitoring with alert badges and clustering for content generation, use the Prompt Tracking page. The Visibility page is the higher-level summary view.

How Visibility Checks Work

When you trigger a Visibility check, the system:

Takes your project's prompt library (the query bank generated during onboarding, or expanded later via the Prompt Library Discover tab)
Sends those prompts to all tracked AI platforms in parallel — ChatGPT, Google AI Overviews, Google AI Mode, Gemini, Perplexity, Copilot, Claude, Grok, DeepSeek, and Meta AI
Parses each response for brand mentions, mention type, sentiment, and competitor co-mentions
Runs a context-health judgment to detect off-topic / wrong-industry / wrong-language answers (LLM-judged, see below)
Calculates platform-specific visibility scores and updates your history

A full 8-platform check typically completes in 30–90 seconds.

Key Metrics Explained

Visibility Score (0–100) — Combines mention rate, mention type quality, confidence, and sentiment.

Score Range	Interpretation
80–100	Dominant — consistently cited as a top recommendation
60–79	Present — appears regularly but not always prominently
40–59	Marginal — occasional mentions, often not in top positions
0–39	Invisible — rarely or never mentioned in relevant queries

Mention Rate — Percentage of your tracked queries where your brand appears. 14 mentions out of 20 queries = 70% mention rate.

Mention Type — Not all mentions are equal:

Recommended — "I recommend X for this use case" (highest value)
Listed — "Some options include X, Y, and Z" (medium value)
Passing Mention — "X is one company in this space" (low value)
Not Found — Brand not mentioned at all

AI Share of Voice (SoV) — Your brand mentions as a percentage of all brand mentions across AI responses. If you're mentioned 30 times and competitors 70 times, your AI SoV is 30%.

Sentiment — Analyzed as Positive, Neutral, or Negative. Negative sentiment often indicates hallucinations or unfavorable training data associations.

Delta — The change in your visibility score compared to the previous check. Drops >10 points trigger an automatic Alert.

Context Health — A separate LLM-judged signal that flags when AI engines describe your brand incorrectly (wrong industry, wrong language, off-topic answer). Independent of whether you're mentioned — the AI can mention you AND get the context wrong.

Reading the Platform Detail Panel

Click on any platform card to expand three sub-tabs:

Overview — Score summary, mention rate, top queries where you appear, and visibility gaps (queries where you don't appear).

Queries — Each tracked query with its result: mentioned/not mentioned, mention type, snippet (exact AI text), sentiment, and competitor co-mentions. Context-mismatch flags appear inline with affected queries.

History — Score trend chart over time for this platform.

Visibility vs Prompt Tracking — Which to Use When

Use Visibility when	Use Prompt Tracking when
You want a project-wide score and platform trends	You want to act on specific gaps (generate content, fix individual prompts)
You're producing a stakeholder dashboard	You're picking 3–7 prompts to cluster into a Strategy Article
You need cross-platform sentiment summary	You want per-row alert badges (Competitor Intercept / Low Visibility)

Most users use both: Prompt Tracking daily for operational work, Visibility weekly for trend reporting.

Frequently Asked Questions

Q: How often should I run Visibility checks?

Weekly minimum for active monitoring. Pro users can set automated daily or weekly checks via Project Settings.

Q: Why do scores vary between platforms?

Each AI platform has different training data, retrieval mechanisms, and response styles. It's normal to be highly visible on Perplexity but less so on Claude — they index and weight sources differently.

Q: Can I add custom queries to my prompt library?

Yes. Go to Prompt Library → Discover tab to expand from a seed query or add new ones manually. We recommend maintaining 15–30 diverse queries across your key topics for statistically meaningful visibility data. See also the seed-picker workflow for expanding from prompts you already track.

Q: What's the difference between "Not Mentioned" and a Context Mismatch?

"Not Mentioned" = your brand name didn't appear in the answer. Context Mismatch = your brand name DID appear but in the wrong context (wrong industry, wrong language, irrelevant topic). Both surface as separate signals so you can target the right fix.

Q: A scan finished but some prompts are still "checking…"

The scan loop has a 3-attempt retry budget per prompt with exponential backoff. Prompts that fail all 3 retries are marked permanently failed for the session — refresh the page or click Run Scan again to retry from scratch.