The Domains That Dominate AI Citations
Around 30 domains account for two thirds of citations in a topic. Here is what that concentration means for anyone trying to be visible.
The open web has billions of pages. AI answers behave as though it has a few hundred, and that gap is the most useful thing to understand about them.
The concentration
Tracking studies across the major engines land on a similar figure: roughly 30 domains account for about 67 percent of citations within a given topic.
Not 30 percent of the web. Thirty domains. Everything else competes for the remaining third.
What kind of sites they are
Three categories, consistently.
Reference sites, with Wikipedia at the front. Professional networks, with LinkedIn increasingly prominent. And established publishers, which took a substantially larger share after the late 2025 shift toward professionally produced content.
What is largely absent is as instructive: company websites, small blogs, and since 2025, forums. Reddit went from appearing in around 60 percent of ChatGPT responses to roughly 10 in the space of a month.
Why it concentrates
Because the problem the engine is solving is trust at speed. It has to pick sources without being able to evaluate them, so it leans on proxies: long search history, editorial process, a named publisher, consistency over time.
Those proxies describe a small number of sites, and they describe the same small number for everybody. Hence the concentration.
What it means for press
It raises the value of placement on recognised publications and lowers the value of volume.
Twenty articles on sites with no search history is a worse position than three on domains inside the cited set, even though the first number looks better on a slide. For a human reader the difference is modest. For a machine it is close to binary.
It also means tier 1 and tier 2 publications are worth distinguishing with some care, because the cited set is drawn almost entirely from them.
The part worth staying sceptical about
Any published list of top cited domains is a snapshot, and the snapshots move. One licensing agreement reshaped ChatGPT's source mix inside a month.
What is unlikely to change is the concentration itself. Whichever names sit at the top next year, there will not be many of them, and they will be publishers.
Getting started
Publisive places clients in the tier 1 and tier 2 publications these studies keep returning to. We write the piece to your brief, you approve every word, and it stays indexed and findable.
We write and place features in the publications our clients choose. About us
Frequently asked questions
01Why do AI citations concentrate in so few domains?
Because an engine has to pick sources without being able to evaluate them, so it leans on proxies: long search history, editorial process, a named publisher, consistency over time. Those proxies describe a small number of sites, and the same ones for everybody.
02Is more coverage better than better coverage?
No. Twenty articles on sites with no search history is a weaker position than three on domains inside the cited set. For a human reader the difference is modest; for a machine it is close to binary.
03Will the top domains change?
The names will. One licensing agreement reshaped ChatGPT's source mix in about a month. What is unlikely to change is the concentration itself, or that the top of the list will be publishers.
