verithia
measurement tool

Nobody owns the AI answer for your category. The most-cited domain holds ten percent.

The fear behind every GEO pitch is that the incumbents own AI search the way they own Google, and a newcomer cannot break in. We measured it. Across 20 categories, Google's top spot has a clear landlord and the AI answer has none. The single most-cited domain holds about a tenth of the citations, and no domain owns a majority of anything.

Original study · measured 2026-08-10

Every pitch for AI search optimization leans on a fear: that the big brands own the AI answer the same way they own the Google results, so a smaller site is locked out before it starts. It is worth testing rather than assuming, because the answer decides whether the work is worth doing at all.

We ran citation_concentration across 20 category buyer queries, measuring, for each, how concentrated the AI citations are: how many distinct domains get cited, and how large a share the single most-cited domain holds. Then we put it next to the Google result for the same queries. The two surfaces could not be more different.

The most-cited domain holds about 10%

Within a category, the AI answer is spread thin. On Gemini, the single most-cited domain averaged a 10% share of the citations, in a tight band from 8% to 12%. Across all 20 categories, not one had a domain holding 50% or more. The mean category drew citations from 18 distinct domains. There is no dominant source. The "leader" in a category is cited a little more than the eighteen others, and that is the whole of its lead.

Google has a landlord. The AI answer does not.

Put the two channels side by side and the contrast is the finding. On Google, one domain owns the top of most categories: Reddit is the number one organic result for 12 of the 20. One landlord, 60% of the categories.

On the AI side there is no equivalent. The single most-cited domain is different in almost every category, and the most any domain "led" was YouTube, top-cited in 5 of 20, and even there it held only about a tenth of the citations. Google concentrates onto an incumbent you cannot outrank. The AI answer refuses to concentrate at all.

Platforms have presence, not ownership

The one thing that carries across categories is the usual platforms, but they do not own anything. YouTube appears in the AI citations of 14 of 20 categories, Reddit in 13, Medium in 8. They are everywhere. Yet in any single category their share sits around that same 10%. They show up broadly and win nothing decisively, which is a very different thing from Reddit sitting at Google's number one across the board.

The citation universe is category-local

Here is the number that should change how you think about your competition: of the 302 distinct domains cited across all 20 categories, 278 of them, 92%, appear in only one category. The domain cited for "best headless cms" is almost never the domain cited for "best vpn." There is no cross-category authority quietly winning everywhere. Your competition for the AI citation is local to your niche, and it is a crowd of small sites, not a national incumbent.

Who "leads," and by how little

category most-cited domain its share distinct domains cited
best ai coding assistant daily.dev 12% 16
best transactional email api mailtrap.io 11% 14
best password manager security.org 11% 19
best headless cms sanity.io 10% 18
best web scraping api scrapfly.io 9% 18
notion alternative thefrontkit.com 9% 21
best crm for startups ahoy.ai 8% 23
best web analytics tool fullstory.com 8% 24
best customer support software youtube.com 8% 26
mean of all 20 varies every time 10% 18.2

Half of the category leaders are small sites you have never heard of, holding a lead of a single percentage point over the pack.

What this means

You cannot be locked out of an answer that nobody owns. On Google, if Reddit or an entrenched brand holds the top of your category, outranking it is a multi-year fight you will probably lose. The AI citation layer is the opposite: the leader in your category holds 10%, no one holds a majority, and the field is 18 small sites deep. That is not a market with a moat around it. It is an open slot.

Run citation_concentration on the cited-domain list for your own queries and read the top-domain share. A high number means someone owns the answer and you have a fight. Across the 20 categories here, the number was 10%, which means the fight is with a crowd, and crowds are beatable.

Method

20 category buyer queries, on 2026-08-10. For each, the AI side is three grounded Gemini runs; citation_concentration is run on the combined cited-domain list, so a domain cited in every run counts more than one cited once, which is what "share" measures (fraction of total citations across the three runs). The Google side is the top organic result per query. Gemini grounding is stochastic, hence three runs. "Distinct domains" counts unique cited domains across the runs; "share" is the most-cited domain's fraction of all citations. This study is Gemini-scoped on the AI side; ChatGPT, which cites from a different index, is measured separately. No scoring, no verdict, raw counts.

A point-in-time measurement. Search results and AI answers change, and grounded models vary between runs, so your own numbers will differ. Verithia measures. The interpretation is yours.

Get a key and run it All tools