Skip to main content

Open in AthenaHQ

app.athenahq.ai/prompts

Purpose

The Prompt Detail page is a dedicated view for analyzing how various AI models answer a single, specific prompt that a customer is tracking. While other pages in Athena aggregate data across hundreds of prompts, this page zooms in to provide granular, row-by-row insight into exactly what models are saying when users ask a specific question. Marketers use this page to evaluate response quality, identify which AI models are mentioning their brand (or their competitors), review the exact sentiment of those mentions, and investigate which third-party sources are being cited. By filtering and exporting this raw data, teams can spot opportunities to improve their AI search presence for a highly targeted query.

What’s on the page

Displays the exact text of the prompt being analyzed. It also houses navigation controls, a button to launch an interactive tour, and a shortcut to view high-level analytics for this specific prompt. A robust contextual toolbar sitting directly above the table. It allows users to search the text of the responses or apply complex filters (like narrowing down to a specific date range, sentiment, or AI model). Users can also manage column visibility, download data, and save their favorite filter combinations here.

Responses Table

The core of the page. It is a paginated, sortable spreadsheet that lists every single time Athena ran this prompt and collected an AI response. Table Columns:
  • Date: The UTC date the response was collected (formatted as MM/DD/YY).
  • Model: The AI model that generated the response (e.g., ChatGPT, Claude, Gemini, Perplexity, AI Overview, Copilot, AI Mode, Grok, DeepSeek).
  • Mentioned: Whether the customer’s brand was mentioned in the response text. Shows a “Yes” badge (indigo), a “No” badge (orange if competitors are mentioned), or a neutral “No” badge.
  • Competitors Mentioned: Logos and names of tracked competitors that appeared in this response. Shows up to 4 logos inline, with a +N badge for any overflow.
  • Sources: Whether the response cited any third-party sources (Yes/No badge).
  • Attributed Citation: Whether the response contained an attributed citation linking back to a source (Yes/No badge). Hidden by default.
  • Market Position: A numeric rank of the customer’s brand among all entities mentioned in the response (where 1 is the top position). Hidden by default.
  • Sentiment: The AI-assessed tone of the response toward the brand (Positive, Neutral, Negative, or N/A for AI Overviews).
  • Content Type: Shows if the response was generated from the original base prompt (“Original”) or a fan-out variation (“Variation”). Hidden by default.
  • Cited: Shows whether any cited source in the response exactly matches the customer’s own website domain (Yes/No badge). Hidden by default.
  • Personas: The specific audience persona segment associated with this response run, if any. Hidden by default.
  • Attributes: Brand attributes detected in the response, displayed as positive (green) or negative (red) badges. Hidden by default.
  • Location: Geographic location(s) associated with this response run. Only visible if the customer’s subscription plan includes the location offering feature.
  • Content: A truncated plain-text preview of the full AI response. Note: For Google AI Overviews, a “View” badge appears instead of raw text.
Shows a summary count (e.g., “Showing X of Y responses”) and a “Load More” button to fetch the next batch of rows via infinite scroll.

Drilldowns

  • Response Drawer (slide-in panel): Clicking any row in the responses table opens a slide-out drawer on the right side of the screen. This drawer reveals the full, uncut AI response text (with brand/competitor highlights), a list of cited sources with favicons, competitor details, assigned brand attributes, and the response’s sentiment. It includes navigation arrows to easily jump to the next or previous response row without closing the drawer.
  • Analytics Button: Clicking the “Analytics” button in the top right takes the user directly to the Olympus dashboard (/olympus), with a filter pre-applied so they only see charts and metrics for this specific prompt.
  • Back Arrow: Clicking the back arrow in the top left returns the user to the exact page they came from (e.g., the main Prompts list, Action Center, or Sources table) while preserving the state of that page (like keeping a specific row expanded or preserving a selected tab).

Export Dialog

A pop-up modal triggered by the Download button. It lets customers choose how many response rows they want to export into a CSV file, complete with quick-select increments and an “Export all” option.

Columns Panel

A popover menu that appears when clicking the “Columns” button. It lists every available table column, allowing users to toggle checkboxes to show/hide them and drag items to reorder how they appear in the table.

What you can do here

Header Actions:
  • Go back: Click the back arrow (<-) to return to the referring page with your previous state intact.
  • Copy prompt text: Click directly on the large prompt text in the header to copy it to your clipboard. A “Copied!” tooltip will appear.
  • Start tour: Click the small info icon () next to the prompt text to launch an interactive guided tour of the page.
  • View Analytics: Click the “Analytics” button to jump to the Olympus dashboard pre-filtered to this prompt.
Filtering and View Actions:
  • Add Filter: Click the “Filter…” button or the + icon to open the filter menu and narrow down responses.
  • Clear filters: Click the reset/clear option to remove all active filters and return to the default view.
  • Saved Views: Inside the filter bar, use the Saved Views dropdown to Save View (save current filters as a preset), Load View (apply a saved preset), or Delete View (remove a saved preset, this is a destructive action).
  • Customize Columns: Click the “Columns” button to open the popover where you can toggle column visibility or click and drag the handle next to a column name to reorder it.
Table Actions:
  • Sort columns: Click a sortable column header (like Date or Sentiment) to reorder the table ascending or descending. This triggers a server-side fetch to sort the entire dataset, not just the visible rows.
  • Resize columns: Click and drag the invisible resize handle on the right edge of any column header to make it wider or narrower. Double-clicking the handle auto-fits the column to its content.
  • Reorder columns inline: Click and drag a column header left or right to move it to a new position directly in the table.
  • Load More: Scroll to the bottom of the table or click the “Load More” button in the footer to load the next 40 responses.
Export Actions:
  • Open Export Dialog: Click the “Download” button in the filter bar.
  • Adjust export limit: Inside the dialog, use the + and - buttons to adjust the export limit in increments of 1,000.
  • Quick select limits: Click the 1,000, 5,000, or 10,000 shortcut buttons to jump to those counts.
  • Reset to 1,000: Click the small reset link to revert the count to the default 1,000.
  • Export all: Click the “Export all” link to set the count to the total number of matching responses.
  • Download CSV: Click the primary “Export” button to generate and download the file. A toast notification will track the progress.
  • Cancel Export: Click “Cancel” to close the dialog without downloading.
Mobile-Only Actions:
  • On smaller screens, the Analytics button, Download button, and Columns toggle disappear from the main layout and are instead tucked inside a trailing overflow menu (the three dots icon) on the right side of the filter bar.

Data shown

  • AI Responses: The actual text generated by various AI models when they were asked this specific prompt, collected either through manual runs or automated schedules.
  • Mentions & Competitors: Data extracted automatically by Athena analyzing the AI’s text to detect references to your brand or your configured competitors.
  • Citations: Links and sources that the AI model referenced in its response.
  • Sentiment & Attributes: AI-driven analysis of the response text to determine its tone and assign any brand-specific attributes you are tracking.
  • Configured Metadata: Information tied to the prompt itself, such as the geographic locations or audience personas it was assigned to target.
  • Competitor Logos: Visual logos fetched dynamically based on the competitor’s domain name.

Common workflows

1. Review responses for a specific prompt
  1. Navigate to the Prompts page and click on a prompt row.
  2. You’ll land on the Prompt Detail page showing all historical responses for that prompt.
  3. Use the “Model” or “Date Range” filter to narrow the results if you are looking for specific trends.
  4. Click any row to open the Response Drawer and read the full AI text, checking for brand sentiment and competitor mentions.
  5. Use your keyboard’s up/down arrow keys or the drawer’s navigation arrows to quickly move through the responses one by one.
2. Filter and export responses to CSV
  1. Apply desired filters in the toolbar (e.g., set Model to “ChatGPT” and Mentioned to “Yes”).
  2. Click the “Download” button in the filter bar.
  3. In the Export dialog, set your desired number of responses using the quick-select buttons, or click “Export all”.
  4. Click “Export”. A CSV file will download automatically, and a progress toast will appear in the bottom corner of your screen.
3. Navigate to Olympus analytics for this prompt
  1. While viewing the Prompt Detail page, click the “Analytics” button in the top right of the page header.
  2. The app will automatically apply a prompt-specific filter and seamlessly navigate you to the Olympus analytics dashboard.
  3. Olympus will load, pre-filtered to only show high-level charts and metrics for this specific prompt.
4. Customize table columns
  1. Click the “Columns” button in the filter bar.
  2. In the popover menu, check or uncheck boxes to hide columns you don’t need or reveal hidden ones (like Personas or Market Position).
  3. Click and hold the drag handle (six dots) next to a column name to drag it higher or lower in the list, reordering it in the table.
  4. Close the popover. Your preferences are instantly saved to your browser’s local storage.
5. Go back to the referring page with state preserved
  1. Click the back arrow (<-) located in the page header.
  2. The app reads invisible “breadcrumb” markers in your URL (like from, returnTo, expandedRow, or tab).
  3. You are navigated precisely back to the page you came from (e.g., a specific tab in the Action Center), and the app restores your previous state, such as keeping a specific row expanded.

Empty, loading, and error states

  • Empty: If you apply filters that yield zero results (or if a prompt hasn’t been run yet), the table body will display a simple “No responses found.” message.
  • Loading: When first opening the page or applying a new filter, an animated loader spinner appears in the center of the table body while data is fetched. Additionally, while competitor logos are fetching, you’ll see a shimmering placeholder skeleton where the image will appear.
  • Error: If the prompt ID in the URL is invalid, the prompt has been permanently deleted, or the user lacks permission to view it, they will be met with a full-screen 404 “This content does not exist” error page.
  • Linked from:
    • The main Prompts list (/prompts), by clicking any prompt row.
    • The global Responses page (/responses), which can link directly to a filtered view here.
    • The Olympus dashboard (/olympus), via drilling down into expanded rows in the charts.
    • The Sources page (/sources), which links here to show prompt details for a specific cited source.
  • Links to:
    • The Olympus dashboard (/olympus), via the Analytics button (with a prompt filter pre-applied).
    • The default referring page via the Back arrow (this is typically /prompts, but could be /content, /content/outline, /sources, or /olympus depending on where the user started their journey).

Common support questions

Why can’t I see the Location column? The Location column is entirely hidden unless your subscription plan includes the location tracking feature. If your account isn’t configured for regional AI search tracking, this column simply won’t appear as an option. Why does the Content column say “View” instead of showing a text preview for some rows? Responses generated by Google AI Overviews do not return raw, easily readable text the way ChatGPT or Claude do. Because the raw data isn’t easily formatted for a tiny table cell, Athena shows a “View” badge. You can click the row to open the drawer and see the full details. Why did my sorting take a moment to update? When you click a column header to sort (like sorting by Sentiment or Date), Athena doesn’t just shuffle the 40 rows you are currently looking at on your screen. It asks the server to re-sort your entire historical dataset for that prompt and fetch a fresh page of results. This ensures your data is perfectly accurate, but it means a brief loading pause. How do I see more columns like Market Position, Cited, or Personas? To keep the table from feeling cluttered, several advanced columns are hidden by default. You can reveal them by clicking the “Columns” button in the filter bar above the table and checking the boxes next to the data points you want to see. Why didn’t the back button take me to the main Prompts page? The back arrow is “smart.” It remembers exactly where you came from by looking at parameters in your URL. If you navigated here from the Action Center or the Sources table, clicking the back arrow will return you to that specific tool (with your previous tabs and rows still open) rather than forcing you to the generic Prompts list.