Technical Analysis

Semantic Map

Visualize the topical relationships between your content and AI Overview citations using AI-powered analysis.

Overview

The Semantic Map uses Google's Gemini model to analyze the semantic relationships between your tracked prompts, the content cited in AI Overviews, and your competitors' content. The result is a visual map of topic clusters that reveals where your content has strong coverage and where gaps exist.

How It Works

When you run a Semantic Map analysis, Gemini processes the text of each prompt and the corresponding AI Overview response content. It identifies thematic clusters by grouping related prompts by topic, then maps which domains are cited within each cluster. This gives you a bird's-eye view of your topical authority relative to competitors.

Reading the Map

The Semantic Map displays topic clusters as distinct groups. Your domain is highlighted within each cluster so you can quickly see where you appear. Gap areas — clusters where your domain is absent — are clearly indicated.

• Larger clusters represent topics with more associated prompts
• Coverage % shows your citation rate within each cluster
• Highlighted nodes represent your domain's presence
• Gray areas indicate topics where you have no citations

Topic Clusters

Each topic cluster includes:

• Cluster name — a descriptive label generated by Gemini
• Related prompts — the prompts grouped into this topic
• Coverage % — how often your domain is cited within the cluster
• Top competing domains — other domains frequently cited in this topic
• Opportunity score — a calculated score based on competitor presence vs. your coverage

Content Opportunities

Clusters with low coverage but high competitor presence represent your biggest content opportunities. These are topics where AI Overviews actively cite sources, but your domain isn't among them. Prioritize creating or improving content for these clusters to expand your AI visibility.

Requirements

The Semantic Map analysis requires a Gemini API key. Analysis runs on-demand and typically takes 30–60 seconds depending on the number of tracked prompts.