Reading the dashboard

The Overview page is the answer to one question: when AI assistants describe my category, am I in the answer?

AI Visibility Score

The hero KPI on Overview. A composite 0–100 score, weighted from four sub-metrics:

Mention rate (40%) — % of responses that mention your brand or any alias.
Citation rate (30%) — % of responses that cite a URL on one of your domains.
Recommend rate (20%) — % of responses that frame your brand as a top pick or recommendation.
Sentiment health (10%) — average sentiment of replies that mention you, mapped from −1..+1 to 0..100.

A score of 100 means the AI consistently mentions, cites, and recommends you with positive framing. A score of 0 means the AI never references you. Most brands land somewhere in 25–60 on first run; the goal is to move the number up over time by closing specific gaps surfaced in Recommendations.

Trend chart

Your visibility score across the last six months, run-by-run. Includes a faint cohort line for context — that's either your real peer cohort (other customers in the same vertical, anonymized) or a synthetic baseline, depending on your data-sharing settings.

Platform breakdown

Same visibility score, broken out by AI platform. Useful for spotting platform-specific weaknesses: e.g. a brand with strong score on ChatGPT but weak on Perplexity often has thin citation quality (Perplexity weights citations more than ChatGPT does).

Citation mix

Donut chart of where citations are coming from. Buckets:

Owned — citations to one of your domains.
Editorial — citations to high-authority publications (caranddriver.com, nytimes.com, jdpower.com, etc.).
Social — twitter, facebook, instagram, tiktok, linkedin.
Reddit — reddit.com (separated because it dominates AI training data).
YouTube — youtube.com / youtu.be.
Other — everything else.

A healthy mix has 30–50% owned, 20–40% editorial, with the rest spread across social/reddit/other. If “owned” is 2% and “editorial” is 70%, you have a content-coverage problem: the AI knows your category, just not via your site.

Best / Worst prompts

The three most useful prompt-level widgets:

Best prompts — top 5 by mention rate. These are wins to defend.
Worst prompts — bottom 5 where you never appear. High-leverage targets.
Mentioned but not recommended — the gap zone. The AI knows you but doesn't recommend you. Usually a sentiment or proof-points problem, not a coverage problem.

Competitor share-of-voice

Each tracked competitor's mention rate vs. yours, with the delta in percentage points. If a competitor sits at 80% and you sit at 12%, you don't have a brand problem — you have a citation density problem.

Insights

Auto-generated narrative observations about your latest run. Mix of rule-derived (deterministic) and LLM-generated (interpretive) insights. Each insight has a severity: opportunity, warning, or risk. Insights are cheap; click Regenerate if the current ones feel stale.