Citation corpus

Which handful of sites is writing your category's answer

4.43%

of German health answers came from YouTube alone, at organic rank 11

SE Ranking, December 2025, across 465,823 AI Overview citations. Your own numbers replace this one.

Citation Sources answers "can I win this page". This answers the question one level up: who owns the category. A list of forty cited URLs does not tell you that five hosts produce half of every answer in your space, and that concentration is the whole argument for treating AI visibility as an off-page problem.

SE Ranking measured the shape at scale in December 2025 across 465,823 AI Overview citations, and found YouTube supplying 4.43% of German-language health answers while ranking 11th organically. Which is a useful illustration and says nothing about your category, so we compute yours instead of quoting theirs.

Where it lives: The leaderboard on Citations, and the top block of the public citation scan.

What it does

Ranks hosts by share of citations

Each row is one host with its share of your whole corpus, rounded. The headline it produces reads like "five domains produce 40% of your category's answers", which is the sentence that changes what someone does next.

Separates pages from citations

A host can own one page cited for eight questions, or eight pages cited once each. Those are different problems and the table counts both, alongside how many distinct questions the host was cited for.

Reports each host's dominant shape

Directory, community, listicle, competitor domain. When no shape holds a majority the row says "mixed" rather than guessing, because calling a host a directory when half its pages are blog posts sends you at the wrong fix.

How it works

  • Computed entirely from rows Citation Sources already produced. No extra query and no extra fetch.
  • Present, absent, and unknown page counts are carried per host, so a domain we have never opened is never rendered as one you are missing from.
  • Each host keeps the date it was last cited, which is how a fading source is told apart from a live one.

What it does not do

  • Share is share of your corpus, not of the whole internet. A narrow question library produces a narrow leaderboard.
  • Subdomains roll up to the host, so a company blog and its docs site count as one property.
  • Concentration is a finding, not an instruction. Some of the top hosts (Wikipedia, G2) are places writing cannot get you into.

Which plans have it

GEO score + cited-thread opportunities. On every plan, and the first five hosts are visible in the free public scan.

GEO score + cited-thread opportunities by plan
BasicGrowthAgency
IncludedIncludedIncluded

Read the full plan comparison.

Try it free

Run the leaderboard on any domain, free, with an email.

AI Citation Source Radar

Live demo

See Domain leaderboard on a real account, with no signup and nothing to configure.

Open the demo

Frequently asked questions

What is a good share to hold on my own domain?

There is no benchmark worth quoting. What matters is the direction over a quarter, which is why the plans default to a three-month cycle: search-grounded engines shift in 30 to 60 days and a single month cannot separate a lift from noise.

Why is a competitor above me when I outrank them in Google?

That gap is common enough that we measure it separately. See the rank overlap check.

Can I see the leaderboard for a domain that is not mine?

Yes. The public scan takes any domain and a category description, and the free Citation Source Radar does the same thing without a login.

Related features

Every feature we publish

Run it on your own product

Start with the free visibility check to see whether the engines name you today, then run the scan that shows which pages they used instead.