When you ask ChatGPT, Perplexity, or Google AI Mode a question, the answer often comes with links. Those links are citations — the sources the AI used (or wants you to trust) when forming its response.
For brands, citations are one of the clearest signals of AI visibility. Being mentioned is good. Being cited with a link is better. Being cited first is best.
What are citations in AI responses?
A citation is a URL that an AI system references in its answer. Sometimes they appear as numbered footnotes. Sometimes they show up as inline links. On Perplexity and Google AI Overviews, citations are often explicit and structured.
Citations serve two purposes for the AI:
- Credibility — The model shows where information came from.
- Freshness — Links point to sources the model considers current and authoritative.
For your business, citations are traffic and trust. If AI recommends your product and links to your site, users can click through directly — without ever visiting a traditional search results page.
Why citations matter for visibility
AI visibility is not just about being named in a paragraph. Citations reveal which domains AI systems treat as evidence.
That matters because:
- Citation = recommendation with proof. A brand mention might be incidental. A citation suggests the AI considered that source relevant to the answer.
- Position matters. The first citation gets more attention than the fifth. In many AI interfaces, top sources are visually prominent.
- Competitor citations are competitive intelligence. If a rival's blog, docs, or landing page is cited more often than yours for the same prompts, you have a content gap — not just a ranking gap.
- Third-party citations shape the narrative. Reviews, news, forums, and directories often get cited even when your own site does not. That influences how AI describes your category.
How Avrae tracks citations
Avrae monitors your brand by querying AI systems with the prompts you care about — the same questions your customers ask when researching products and services.
For each response, we extract URLs using the best available method for each AI system:
- Structured sources from Perplexity, Gemini, and Google AI Overviews when the API returns them directly.
- Regex fallback for other systems when links appear in plain text.
Each citation is then classified:
| Classification | Meaning |
|---|---|
| Own domain | A page on your website was cited |
| Competitor | A tracked competitor's domain was cited |
| Third-party | Another site (news, forums, directories, etc.) |
| Source type | Corporate, organic, news, forum — based on domain patterns |
| Page type | Landing page, article, product page, documentation, and more |
We also record citation position (first, second, third link in the response) and which AI system produced the citation, so you can compare behavior across ChatGPT, Perplexity, Gemini, Claude, Grok, and Google AI Mode.
Reading the citations dashboard
The Citations page in Avrae gives you an org-wide view of every URL cited across your tracked prompts.
Key metrics at the top:
- Total Citations — All unique citation events we detected.
- Your Domain — How often your site was linked.
- Competitor Citations — How often rivals were linked instead.
- Your Citation Rate — Your domain citations as a percentage of total citations.
The table columns:
- URL — The cited page, with favicon and title when available.
- Source — Own domain, competitor, or third-party category.
- Type — Page type (article, landing page, product page, etc.).
- AI System — Which model cited this URL.
- Pos. — Citation position in the response (#1, #2, …).
- First Seen — When we first detected this citation.
Filters help you dig in:
- Search by URL or domain
- Filter by date range
- Toggle between all citations, own domain, competitor, or third-party
- Filter by AI system, source type, or specific prompt
Use these together to answer questions like: "Which competitor pages does Perplexity cite for our top commercial prompts?" or "Are we getting cited for our product pages or only our homepage?"
Actionable insights from citation data
Citation data is most useful when it drives decisions:
If competitors are cited more often
- Study which pages get cited (comparison articles, pricing pages, docs, case studies).
- Create or improve content that directly answers the prompts where you lose.
- Ensure key pages are crawlable, clear, and structured for AI extraction.
If third-party sites dominate
- Invest in PR, reviews, and directory presence on domains AI already trusts.
- Participate in forums and communities that appear in your citation data.
- Build partnerships with publications cited in your category.
If you're cited but not mentioned by name
- Your content is doing work, but brand association may be weak.
- Strengthen on-page branding, author bios, and structured data.
- Consider an LLMs.txt file to help AI systems understand your site.
If citation rates differ by AI system
- Each model has different source preferences. Perplexity may favor news; ChatGPT may favor authoritative docs.
- Tailor content format and distribution to the systems that matter most for your audience.
Citations vs. brand mentions
Avrae tracks both mentions (your brand name appearing in text) and citations (URLs being linked). They tell different stories:
- A mention without a citation means AI knows your brand but didn't point users to your site.
- A citation without a strong mention might mean your content was used as background evidence.
- The best outcome is both: your brand named clearly and your URL cited prominently.
Monitor both over time. As AI search grows, citation share may become as important as traditional share of voice.
Citations are the receipts AI systems leave behind. Avrae collects them automatically so you can see who gets linked, for which questions, and on which platforms — then act on that intelligence before your competitors do.
Run a free visibility report to see how your brand appears across AI systems, or explore the Citations dashboard on a paid plan for ongoing tracking.