Skip to main content
Sources answers one question: when AI search engines write about your market, what websites do they read? When ChatGPT, Perplexity, Gemini, Claude, Copilot, Google AI Overview, AI Mode, Grok, or DeepSeek answer a tracked prompt, they almost always cite one or more web pages as evidence. Sources collects every one of those citations across every prompt you track, rolls them up by website (and by individual URL), and shows you which pages are influencing AI’s answers about your brand, your competitors, and your category. It’s one of the highest-leverage pages in Athena, because the sites that show up here are the sites you need to be present on. Yet many customers underuse it - they look at Olympus and stop. Sources is where you turn AI visibility insights into concrete content, outreach, and SEO actions. You’ll find Sources in the left sidebar.

​
Why Sources matters

Three concrete things you can do with Sources that you can’t do anywhere else in Athena:
  1. Find where to publish or pitch. The third-party sites that show up at the top of your Sources list (Reddit threads, Forbes articles, Wikipedia pages, niche industry blogs, YouTube videos) are the sites AI is already trusting to answer questions about your market. Getting your brand mentioned in those places is one of the most direct ways to influence AI answers.
  2. Spot content gaps. If a domain drives a lot of competitor citations but rarely cites you, that’s a measurable, named gap to close - either by getting your own content mentioned there, or by publishing better content on a similar topic on your own site.
  3. Audit your own performance as a source. Switch the source type to Owned for your domain and you can see exactly which of your pages are being cited by AI, how often, and alongside which competitors. This is the closest thing Athena has to a direct AI-SEO performance report for your own content.

​
What you can do here

​
Two ways to look at the data

A toggle at the top of the page switches between:
  • Domain view - one row per website (e.g. reddit.com, forbes.com, yourcompany.com). Best for spotting which sites matter most.
  • Page view - one row per individual URL. Best for drilling into exactly which articles, threads, or videos are doing the work.
A second toggle splits the table into:
  • All - every domain or URL.
  • Social - social platforms only (YouTube, Reddit, X, LinkedIn, Facebook, TikTok, Quora, Substack, etc.). When you’re on the Social tab, platform chips appear so you can narrow to one platform at a time.

​
Reading the columns

The main columns work the same way in both views:
  • Citation % - the share of AI responses (under your current filters) that include at least one citation to this domain or URL. Higher = more influential.
  • Responses Influenced - the raw count of distinct AI responses that cited this source.
  • Brand Mention % - of the responses that cite this source, how often your brand is also mentioned in the answer.
  • Competitor Mention % - of the responses that cite this source, how often a tracked competitor is also mentioned.
  • Trend - a small sparkline showing how citation share has moved across your selected date range.
  • Type - the source-type tag (Owned / Competitor / Partner / Third-party). Color-coded throughout the product.
  • Impressions Captured - estimated total AI impressions delivered through citations to this source.
Page view also shows Mentions (the same as Brand Mention but per URL), a stack of Competitor logos for tracked competitors co-mentioned alongside the URL, and an optional First Seen column (turn it on via the Columns menu) showing when the URL first appeared in an AI response.

​
Drilling into a source

Click any row to open a drawer with deeper analytics for that domain or URL:
  • Citation Rate Over Time - a line chart of citation share over your selected date range.
  • URL Breakdown - every URL on the domain that’s been cited, with the same columns.
  • Prompt Breakdown - which of your tracked prompts produced these citations.
  • Creator Breakdown and Timestamp Insights - extra tabs that appear only for YouTube, surfacing video creators and the specific timestamps AI cites.
From the URL drawer you can navigate up the URL stack (click the domain or path segments in the header) and export the breakdown to CSV.

​
Filtering and saved views

Above the table is a filter row with date range, models, prompts, competitors, personas, locations, prompt tags, prompt status, source type, “mentioned,” and a free-text search box. Any combination can be saved as a Saved View for one-click recall later. Saved views persist per website. If no date range is set, Sources defaults to the last 7 days.

​
Tagging sources

Every domain and URL is automatically classified as Owned, Competitor, Partner, or Third-party based on your website, your tracked competitors, and a default list. If something is misclassified - or if you want to flag a partner site or industry publication as something more specific - you can override it:
  • Single row: click the … menu on any row and choose Edit Source Type.
  • Multiple rows (Page view only): check the rows you want, then use Assign Type in the bulk command bar that appears at the bottom of the screen.
Re-tagging updates the badge immediately and flows through to filtering and reporting.

​
Rescanning brand and competitor mentions

For each URL that AI cites, Athena scans the page to see whether your brand and tracked competitors are mentioned in the surrounding content. This is what powers the Brand Mention % and Competitor Mention % columns. You can refresh that data:
  • One URL at a time: click the rescan icon on the row (Page view only).
  • In bulk: click the Rescan column header in Page view to open the bulk rescan dialog. Pick a scope (Never scanned, Not mentioned, No competitors, or All), choose how many URLs to scan (up to 10,000 per run), and start the job. Each batch consumes about 1 credit per 10 URLs scanned, and the dialog shows you both the cost and your remaining balance before you confirm.

​
Exporting

Click the Download icon at the top right of the table to open the Export Sources dialog. You can export 1,000 / 5,000 / All rows, or any custom count. The CSV downloads automatically. Each drawer also has its own download button for exporting the URL breakdown for a single domain.

​
Suggested workflows

Find the top sites AI cites about your market
  1. Open Sources (Domain view, All tab).
  2. Sort by Citation % descending.
  3. Skim the top 10-20 domains. Note which are owned, competitor, or third-party.
  4. Click the most influential third-party domain to see which prompts it’s appearing on.
Find content gaps (sites that talk about competitors but not you)
  1. Open Sources in Domain view.
  2. Sort by Competitor Mention % descending.
  3. Look for rows where Brand Mention % is low. Those are the highest-leverage domains to target.
Audit your own AI-SEO performance
  1. Filter Source Type to Owned.
  2. Switch to Page view.
  3. Sort by Responses Influenced.
  4. These are your top AI-SEO performers - the pages on your site that AI is actually citing.
Plan social outreach
  1. Click the Social tab.
  2. Pick a platform chip (Reddit, YouTube, etc.).
  3. Sort by Responses Influenced. The top rows are the specific threads and videos AI relies on most for your market.
  4. Click a row to read the underlying AI responses and decide where to engage.
Re-tag a partner or industry site
  1. Switch to Domain view.
  2. Search for the domain.
  3. Use the row’s … menu > Edit Source Type > choose Partner.
  4. Reporting and color-coding update everywhere immediately.

​
Frequently asked questions

Where does the citation data come from? Athena runs every prompt you track against the major AI search engines (ChatGPT, Perplexity, Gemini, Claude, Copilot, Google AI Overview, AI Mode, Grok, and DeepSeek), captures their responses, and extracts every domain and URL each response cites. Sources is built entirely from that data - no third-party SEO tool is feeding it. What does “Citation %” actually count? It’s the share of all AI responses captured under your current filters that include at least one citation to this domain (or URL, in Page view). Why is a YouTube URL stripping the ?v=... part on other sites but keeping it on YouTube? For most domains, Athena normalizes URLs by stripping query parameters so duplicates collapse into one row. For social platforms (especially YouTube /watch and /shorts URLs), the query parameters are preserved so each video tracks separately. How do I see when a URL was first cited? Switch to Page view, click Columns, and turn on First Seen. It’s hidden by default. Why don’t I see the Domain / Page or All / Social toggles? They’re hidden on narrow screens (mobile or very small browser windows). Widen the window or use a desktop browser. I re-tagged a domain but other pages still show the old type. What happened? The badge updates immediately in Sources. Other reports run on a server-side recompute and may take a moment to reflect the change. If it still looks wrong after a full reload, contact support. Why is the bulk Rescan button disabled? Either another bulk rescan is already in progress for this website, or you don’t have enough credits. The dialog shows the credits needed (about 1 per 10 URLs) and your remaining balance. Can I export the full list? Yes - click the Download icon at the top right of the table and choose All in the Export Sources dialog. Drawers also have their own per-domain export. Do saved views sync across users? Saved views are saved per website, so any user with access to that website’s Sources page can load them. Sort order, column widths, and column order are stored in your browser’s local storage and stay on the device you set them on. Why is the same source listed twice with a slightly different URL? The page treats every distinct URL Athena has captured as its own row. Common reasons for near-duplicates: trailing slashes, anchors (#section), or social-platform query parameters that are intentionally preserved. Use the Clear URL Param action inside the URL drawer to roll a suffixed URL back to its base.