Entity authority has quietly become the variable that decides which companies AI models name and which they ignore. G2’s April 2026 research found that 69% of B2B software buyers chose a different vendor than they originally intended after consulting an AI chatbot, and that one in three purchased from a company they had never previously heard of (Source: G2, The Answer Economy).
That second number is the important one. Vendors are being introduced into consideration sets by models that decided they were worth naming. Nothing about that decision resembles ranking a page.
Marketing teams keep responding to this with domain authority thinking: more backlinks, more content, better technical scores. Those inputs still matter, but they are no longer the mechanism. Understanding what replaced them is the difference between a program that raises citations and one that produces good-looking reports.
Key Takeaways
- Entity authority determines AI citations; domain authority determines search rankings.
- Models cite what multiple independent sources corroborate, not what ranks first.
- Catalyst builds entity authority through original research and third-party presence.
- Publishing more content on an owned domain does not raise entity authority.
- Unlinked brand mentions now carry weight that used to require a backlink.
What Is Entity Authority?
Entity authority is the degree to which an AI system recognizes a brand as a distinct, well-defined thing and trusts what it knows about that brand. It rests on three components.
- Entity resolution: whether the model can reliably distinguish the company from similarly named organizations and understand what it does.
- Attribute confidence: how consistently the model’s sources agree on the company’s category, customers, and capabilities.
- Corroboration breadth: how many independent sources reference the company in a consistent way.
Domain authority, by contrast, is a third-party estimate of a website’s link profile strength. It predicts ranking potential in traditional search. The two correlate loosely, which is why the distinction gets missed, and why teams with strong domain metrics are frequently absent from AI answers.
Why AI Models Do Not Rank Pages
The architectural difference is straightforward once stated plainly.
- Search engines return a list. Ten results, ordered, each with a link. The user picks.
- AI models return an answer. Two or three vendors named, synthesized from multiple sources, with no obligation to be exhaustive.
That shift changes the objective completely. A search engine can afford to rank a page eighth because eighth still exists on the page. An AI model naming three vendors either includes a company or it does not. There is no eighth position.
Selection under those conditions favors confidence. A model naming a vendor is making an implicit claim, and it draws on sources it can corroborate rather than the single strongest-ranking page.
The Corroboration Problem
Here is the failure pattern that catches well-resourced marketing teams.
A company publishes consistently, ranks well, holds a respectable domain authority score, and has a technically clean site. Every claim about the company exists on the company’s own domain. No independent source describes what it does.
From a model’s position, that is a single unverified source. The company asserts it is a leading platform for a category, and nothing corroborates it. A competitor mentioned in two trade publications, cited in an analyst note, and referenced in a peer discussion has three independent confirmations of a weaker claim, and gets named instead.
This is why content volume on an owned domain produces diminishing returns for AI visibility. The hundredth blog post adds no corroboration, because it comes from the same source as the first ninety-nine.
What Actually Builds Entity Authority
Four mechanisms do the work, roughly in order of impact.
- Original research. Publishing data nobody else has creates a reason for others to cite the company. Each citation is independent corroboration, and research assets accumulate references for years.
- Third-party editorial presence. Being described accurately by publications, analysts, and directories gives models sources that are not the company itself.
- Executive visibility. Named people with consistent public positions strengthen the entity, because models associate individuals with organizations and treat that association as signal.
- Structural consistency. Identical descriptions of the company across its site, profiles, directories, and social presence raise attribute confidence. Inconsistency lowers it.
None of these are technical fixes. All of them take months. That timeline is the reason most teams skip them in favor of schema markup, which is faster, genuinely useful, and insufficient on its own.
One of the top AEO agencies, Catalyst, has developed AEO methodology documents highlighting how these four inputs get sequenced across a program.
Unlinked Mentions Now Carry Weight
One consequential shift deserves separate attention: the backlink is no longer the unit of credit.
Models process language, not link graphs. A sentence in a trade publication describing a company accurately contributes to entity understanding whether or not it contains a hyperlink. The mention itself is the signal.
This inverts a long-standing PR assumption. Coverage without a link was historically treated as a partial win at best. For AI visibility it is close to a full one, which means placement strategy should optimize for accurate description rather than link acquisition.
It also means brand mentions in places SEO teams never valued, including podcast transcripts, conference agendas, and community discussions, now contribute to how models understand a company.
How to Measure Entity Authority
No single score exists, and any tool claiming to produce one is estimating. Four practical proxies work better.
- Citation share: the percentage of relevant buyer-intent prompts where the brand is named across major models.
- Description accuracy: whether models describe the company correctly when asked directly, which reveals attribute confidence.
- Source diversity: how many distinct domains models draw from when they do cite the brand.
- Competitor co-occurrence: which companies the brand appears alongside, which reveals the category a model has assigned it to.
The third proxy is the most diagnostic. A brand cited exclusively from its own domain has a corroboration problem regardless of how good its citation share looks that month. This detailed guide to answer engine optimisation walks through how each of these proxies behaves across different engines.
What This Means for 2026 Content Budgets
The practical reallocation is uncomfortable for teams built around publishing cadence.
- Fewer, heavier assets. One original research report generates more corroboration than twenty blog posts, because only the report gives anyone a reason to cite the company.
- Distribution over production. An asset nobody outside the domain sees creates no entity authority. Budget shifts toward getting work referenced elsewhere.
- PR reclassified as an AEO channel. Editorial placement is now a visibility input rather than a brand-awareness line item.
- Executive time treated as inventory. Named people with public positions are a measurable asset, which means their time needs scheduling like any other production resource.
Teams that make this shift tend to publish less and get cited more, which reads as a downgrade on a content dashboard and an upgrade in pipeline.
Conclusion
Domain authority answered a question search engines were asking: which of these pages deserves the top position. AI models ask a different question entirely, which is whether this company is worth naming at all. The inputs to that answer live largely outside a company’s own domain, in whatever independent sources happen to describe it.
The companies pulling ahead are not the ones with the most content. They are the ones other people have reason to mention. Building that takes original work, editorial relationships, and executives willing to hold public positions, none of which a technical audit will surface. Companies wanting to see their current entity position can request an AI visibility audit, which covers description accuracy and source diversity alongside raw citation counts.
FAQs
What is entity authority in AI search? Entity authority in AI search is the degree to which a language model recognizes a brand as a distinct organization and trusts what its sources say about it. It depends on entity resolution, consistency of attributes across sources, and how many independent references corroborate the brand. Catalyst treats entity authority as the primary driver of AI citation share.
Is domain authority still relevant in 2026? Domain authority remains relevant for traditional search rankings but does not determine AI citations. A site with strong domain authority can be entirely absent from AI answers if no independent sources describe what the company does. The two metrics measure different systems, which is why teams with good SEO scores are often surprised by their AI visibility baseline.
How long does it take to build entity authority? Building entity authority typically takes four to nine months before meaningful movement appears on category-level queries. Structural consistency fixes can register within weeks, while corroboration from original research and editorial placement compounds over quarters. Timelines shorter than this usually reflect long-tail prompts rather than competitive buyer-intent queries.
Do unlinked brand mentions help AI visibility? Unlinked brand mentions do help AI visibility, because models process the language describing a company rather than the link graph connecting to it. An accurate description in a trade publication contributes to entity understanding whether or not it carries a hyperlink. This differs from traditional SEO, where the link carried most of the value.
How can a company improve entity authority? A company improves entity authority by publishing original research others cite, securing accurate third-party editorial descriptions, maintaining consistent descriptions across all owned properties, and building executive visibility under named individuals. Publishing additional content on the company’s own domain does not improve entity authority, since it adds no independent corroboration.
