Skip to main content

Open in AthenaHQ

app.athenahq.ai/sources/visualization

Purpose

The Sources Visualization page provides an interactive chart that maps out exactly where AI models (like ChatGPT and Perplexity) are getting their information when they talk about your brand or your competitors. By visualizing the connections between web sources and brand mentions, you can easily spot which domains are driving the most value for you, uncover sources that are heavily feeding competitor recommendations, and identify content gaps where high-traffic domains mention competitors but leave you out.

What’s on the page

Shows the title “Sources Visualization” along with an info icon. Hovering over the info icon displays the following explanation:
  • More About Source Attribution: This graph shows where LLMs get their information from when they mention your brand and competitors. Thicker connections = more responses where an LLM mentioned the brand (RIGHT) and cited a source (LEFT) that mentions the brand.
  • LLM Citations Deep Dive: When an LLM mentions your brand, it is from finding a source that mentions your brand. This is finding those deep sources, and pointing you to them.
  • Competitor intel: See which sources drive competitor recommendations.
  • Content gaps: Find high-traffic sources that mention competitors but not you.
  • Click any connection to see the actual LLM responses

Top Sources Chart

An interactive visualization (often called a Sankey diagram) that maps the flow of citations.
  • Left-side nodes (Indigo): The web sources (domains) that AI models cited.
  • Right-side nodes: The entities being mentioned by the AI. Your brand appears in green, while tracked competitors appear in orange.
  • Connections (Lines): The flowing lines represent responses where the AI cited the source on the left and mentioned the brand/competitor on the right. Thicker lines mean a higher volume of responses.

What you can do here

  • Magnify the chart: Toggle the Magnify switch above the chart to enable a fisheye effect. When enabled, moving your mouse over the chart smoothly enlarges the nodes and links nearest to your cursor, making it easier to read dense clusters of data.
  • Focus on a specific node: Click on any node (a source on the left or a brand/competitor on the right) to enter focus mode. The chart will filter down to show only the connections flowing into or out of that specific node.
  • Reset the view: When focused on a single node, a Reset view button appears above the chart. Clicking this (or clicking the node again) restores the full chart.
  • View node details (Hover): Hover over any node to see a tooltip displaying the domain/brand name, the total number of citations it has, and its percentage share of the total.
  • View connection details (Hover): Hover over the line connecting a source to a brand. A tooltip will appear showing the exact number of responses making that connection, the percentage share, and a breakdown of the flow (e.g., “① LLM cited [Source] as a source” and “② [Source] & the LLM mention [Brand]”).
  • Drill down into responses (Click): Click on any connection line to drill down. This navigates you to the standard Sources page, automatically applying filters so you can read the actual AI responses that cited that source and mentioned that brand/competitor.
  • Add and adjust filters: Use the Add filter menu to refine the chart by Date Range, Models, Topics & Prompts, Competitors, Personas, Locations, Countries, Prompt Tags, Prompt Status, Mentioned, Source Tag, Attributes, and Brand Identifiers.
  • Manage saved views: Click the Views dropdown to:
    • Save as view: Save your current filter configuration.
    • Load a view: Click an existing view to apply its filters.
    • Edit: (Inside the ⋯ menu) Rename the view or change its icon.
    • Update filters: (Inside the ⋯ menu) Overwrite the saved view with whatever filters are currently active on your screen.
    • Set as default / Remove default: (Inside the ⋯ menu) Choose which view loads automatically when you visit the page.
    • Export as CSV: (Inside the ⋯ menu) Download a CSV file of the prompts that match the view’s filters.
    • Delete: (Inside the ⋯ menu) Permanently remove the view (requires confirmation).
  • Use mobile filters: On smaller screens, click the Filters button to open a drawer containing all available filter fields for touch-friendly editing.

Data shown

The visualization is built from the AI responses collected for your tracked prompts across your selected models (such as ChatGPT, Claude, and Gemini) and date range. It aggregates citation and mention data, mapping the web domains cited by the AI to mentions of your brand and your configured competitors.

Common workflows

Find top sources driving brand mentions
  1. Ensure the chart is showing data for your desired date range and models.
  2. Look at the left side of the chart for the tallest (thickest) source nodes, which indicate the highest citation volume.
  3. Hover over the lines connecting these top sources to your brand’s green node on the right to see exactly how many responses they share.
  4. Click the connection line to drill down and read the specific AI responses.
Investigate competitor-driving sources
  1. Use the filter bar to add a Competitors filter, narrowing down the right-side nodes to specific rivals.
  2. Scan the chart for thick orange lines connecting a left-side source to a competitor node.
  3. Click the connection to view the underlying responses, helping you understand why that source is heavily recommending them.
Focus on a single source or brand node
  1. Click on a specific node in the chart (for example, a high-traffic source domain you want to analyze).
  2. The chart will immediately filter out unrelated data, isolating only the connections flowing in and out of that selected node.
  3. When you are done investigating, click the Reset view button above the chart to return to the global view.
Save a current filter combination as a view
  1. Add and adjust your desired filters (e.g., specific models, a certain competitor, and a custom date range).
  2. Click the Views dropdown in the top left.
  3. Click Save as view, type in a memorable name, and select an icon.
  4. Click Save. The view is now stored in the dropdown for quick access later.

Empty, loading, and error states

  • Loading: While the chart data, website details, or competitor lists are being fetched, a spinning animated loader is displayed in the center of the chart area.
  • Empty: If there is no data available for your current filter combination, if you have no competitors configured, or if the system is waiting for data to process, the chart area displays a dimmed, blurred-out placeholder graphic. This is intentionally blurred to indicate that no real data matches the current criteria.
  • Linked from: The main navigation menu (under the Sources section as the “Sources Visualization” tab), and from the standard Sources page (if a user clicks back after a drilldown).
  • Links to: The standard Sources page (when clicking a chart connection to drill down into the specific AI responses).

Common support questions

What does the thickness of the connection mean? The thickness of a line represents the volume of AI responses. A thicker line means the AI cited that specific source and mentioned that specific brand/competitor together more frequently. Why is the chart blurry or dimmed out? The chart appears blurred and dimmed when there is no data to display for your current filters, or if you don’t have any competitors set up yet. Try broadening your date range or removing some filters. How do I see the actual AI responses? Click on any of the flowing lines connecting a source to a brand. This will take you to the standard Sources page, automatically filtered to show the exact responses that make up that connection. What does the Magnify toggle do? The Magnify toggle turns on a fisheye lens effect. When enabled, the parts of the chart closest to your mouse cursor will smoothly expand, making it much easier to read the labels and see individual lines in areas where the chart is tightly packed.