Across 22,881 AI citations analysed in mid-2026, 34.5 percent came from sites with Domain Authority below 40, which suggests that a substantial share of AI answer sourcing is happening well outside the publications most brands are competing for.
Where the number comes from
The figure was published by Featured on 25 August 2026 alongside the launch of a product called GEO Audit. Their dataset covers 22,881 AI citations across 405 audits run between 2 June and 21 August 2026.
We have no relationship with Featured and have not independently verified the dataset. Read it as their finding rather than ours.
What makes it worth citing anyway is that it arrives with a stated sample size, a stated date range and a stated method, which is more than most numbers in this category come with. A figure you can interrogate is worth more than a figure that sounds better.
Why the result is plausible
Retrieval and ranking are not the same problem.
Classic search ranking rewards authority accumulated over time. Retrieval for a generative answer rewards specificity: a page that addresses the precise question being asked, in language close to the question, in a structure a model can lift from.
A specialist page on a small site can beat a general page on a large one, because the small page is about exactly the thing. That is a different competitive dynamic from the one most PR planning assumes.
None of which means domain authority is irrelevant. It means it is one input among several rather than the entry ticket.
What this changes about budget
Most AI visibility plans we see contain some version of the same line: get covered in three or four major publications this quarter.
Those placements are expensive, slow, competitive, and frequently the hardest possible way to enter a citation set. If a third of citations are coming from below DA 40, the same budget spread across a wider set of specific, relevant, smaller sources may produce more retrievable surface area.
Three practical implications.
Relevance beats prestige more often than the planning assumes. A trade publication read by four thousand people in your exact category can be a better retrieval target than a general business title read by four million.
Your existing long tail may already be doing work. Before commissioning anything, look at which domains are actually cited in answers about your category today. Some of them will surprise you and several will be small.
Cheap does not mean easy. A smaller site is easier to reach and no easier to be genuinely useful in. The bar for being worth citing is unchanged.
How to build a target list from this
The finding is only useful if it changes what you do next, so here is the sequence.
Collect the actual citation set first. Ten to twenty realistic buyer questions for your category, run without naming your brand, across ChatGPT, Claude, Gemini and Perplexity. Record every domain cited.
Sort by how often each domain appears, not by how impressive it is. The frequency ordering and the prestige ordering are usually different, and the frequency one is the one that matters.
Separate the reachable from the aspirational. Some cited domains accept contributed pieces, run practitioner interviews, or maintain directories. Others take a year of relationship building. Both belong on the list and they belong in different columns.
Check for the thing that is already yours. A surprising share of citations in most categories go to review marketplaces, community threads and documentation, none of which are commissioned coverage at all.
The caution that goes with it
A category aggregate cannot tell you what is happening in your category.
34.5 percent across 405 audits spanning many industries is a fact about the dataset. Your own category may skew heavily toward a handful of large publications, or further toward the tail than the average. The number is a reason to go and look at your own citation set rather than a finding about it.
The check takes about twenty minutes. Pull the sources cited in AI answers for ten buyer questions in your category and sort them by size. If the distribution looks nothing like the aggregate, believe your own data.
The question underneath the number
Once you know which domains the engines cite for your category, a second question arrives immediately: how did each of those pages come to exist?
A citation from a journalist who found the story and a citation from an article somebody paid to place look identical in a report. That distinction is becoming more important as source influence becomes targetable, and it is not something a citation count can answer.
FAQs
What percentage of AI citations come from low authority sites?
In data published by Featured in August 2026, 34.5 percent of 22,881 analysed AI citations came from sites with Domain Authority below 40.
Does domain authority still matter for AI visibility?
It appears to matter less than it does for classic search ranking. Retrieval for generative answers rewards specificity and relevance to the question, so a specialist page on a smaller site can outperform a general page on a larger one.
Should I stop pursuing large publications?
No. The finding suggests broadening the target list rather than abandoning it, because prestige placements still carry value that citation counts alone do not capture.
How do I find which sites AI engines cite in my category?
Run ten realistic buyer questions across ChatGPT, Claude, Gemini and Perplexity without naming your brand, and record every domain cited in the answers.
Is a category-wide citation statistic applicable to my brand?
Not directly. An aggregate across many industries is a reason to examine your own citation set, not a description of it.
What sample does the 34.5 percent figure come from?
It comes from 405 audits covering 22,881 AI citations, run between 2 June and 21 August 2026, published by Featured. We have not independently verified the dataset.
ABOUT AXIS SUITE
Axis Suite by TrendAxis is the independent intelligence layer that explains what AI believes about your brand, why it believes it, and what decision that belief ultimately drives.
Most tools count citations. Axis Suite tells you which of those citations are actually evidence about you, how many separate parties they represent, and what would remain if your largest source disappeared.
Run a scan at axissuite.ai or read the methodology in the Proof Center.