AI Citation Source Tracker: Which Sites AI Cites About You

Why AI citation sources are the new backlink graph

When ChatGPT, Gemini, Claude, or Perplexity answer a question about your brand, they don't make the answer up — they ground it on specific external sources. Sometimes those sources appear in a visible sources panel; sometimes they're implicit in the model's training data; sometimes they're surfaced by the engine's live retrieval layer. Either way, the set of sources doing the grounding is the substrate of your AI presence.

Traditional SEO trained us to think about backlinks: which sites link to you. AI citation sources are a different graph: which sites AI cites about you. The two overlap, but only partially. A high-DR backlink on a page that AI engines never read is invisible. A medium-DR mention in a G2 roundup that ChatGPT scans weekly is enormously valuable. Lumos indexes the citation graph AI actually uses — currently 241,000+ citation events across the major engines — so you can see what's doing the work.

What you see inside Lumos

The preview above is Lumos's own citation source tracker, run on our brand across the major engines this week. Signing up runs the same view on your brand.

• Top domains by citation count. The cited domain, the number of times it appeared as a source in the last 7 days, and the engines it appeared in.
• The driving prompts. Each domain is linked to the specific prompts where it surfaced. Knowing that g2.com is your #1 citation source is useful; knowing that it's because of a single "best GEO tools" roundup is actionable.
• Citation type tags. Each citation is tagged as your own property, a third-party review/roundup, a news article, a directory (Crunchbase, LinkedIn), or a community post (Reddit, HN). The mix tells you a lot about your authority shape.
• Week-over-week tracking. You see when a new high-impact domain starts citing you — or stops — so you can react before it shows up in your visibility score.

How to use the citation tracker

As a PR target list. Run the citation tracker on a strong competitor in your category. The domains that cite them most are the domains most likely to cite a new story about you fast — their audience and the engines' grounding both already associate them with the topic. We've seen teams pitch the top five competitor citation domains and land coverage within 60 days.

As an audit of authority shape. A healthy citation mix has roughly 30-40% your own domains, 30-40% third-party reviews and roundups, 10-20% news, and 10-20% communities. If your mix is 90% your own domains, you're a brand that AI knows mostly through your own marketing — fragile to a content change. If it's 90% community posts, you have buzz but no authority. The mix tells you what kind of presence you have.

As an early warning system. When a new domain shows up in your top 10, that's a signal — usually a new review, a roundup, or a coverage event. Sometimes it's negative (a comparison post that rates you poorly) and you want to respond fast. Lumos alerts you on every new domain entering the top 10.

As a backwards content brief. Look at the pages on your own domain that AI cites most. Those are your strongest content assets. Refresh them quarterly and you protect a chunk of your AI visibility.

Common mistakes when reading citation sources

Treating citation count and citation impact as the same. A citation in a Crunchbase profile is a different thing from a citation in a G2 review or a TechCrunch story. Crunchbase moves on its own pace; G2 reflects buyer trust; TechCrunch grows category awareness. Count is a starting point — read the type.

Optimizing for citations from your own domain. Self-citations are real and helpful, but they top out. Past a certain point, every additional citation from your own pages adds less than a fresh third-party mention. Most brands hit this ceiling around 35% self-citation share.

Ignoring engine differences. ChatGPT and Claude often agree on sources because they share grounding patterns. Perplexity and Gemini diverge because they ground on different web slices. A domain that's invisible to one engine but loud in another tells you which engine's retrieval favors that source.

Chasing the top domain instead of the gap. The single biggest mistake is over-optimizing the domain that already cites you most. The leverage is in the gap — domains that cite your competitors but not you. That's what the Citation Gap Finder is for.

Next steps

• Run the citation tracker on your brand: sign up and see your top sources in minutes.
• Find your citation gaps: Citation Gap Finder.
• See where you rank when cited: AI Citation Position.
• Read the strategy: Why GEO is a competitive advantage.