Why AI Engines Cite Some Brands and Ignore Others

AI engines cite brands they can confidently identify, that other credible sources back up, and that give them something accurate and quotable to repeat. Brands missing any one of those three are easy to leave out, and most are missing at least one. Citation isn’t luck. It’s the product of signals you can either earn or neglect.

Ask the same question of ChatGPT or Perplexity a few times and you’ll notice the same handful of names keep coming back. That consistency is the tell: the engine isn’t guessing, it’s resolving to sources it trusts. Understanding why those names win, and yours might not, is the whole game.

So why does the AI keep naming the same few?

Three forces decide it, and they reinforce each other.

The first is identity. An engine can only cite what it can cleanly resolve. If your business has a consistent, machine-readable presence, with the same name, the same description and the same entity across your site and the wider web, the model knows exactly who you are. If your identity is fuzzy or collides with something similarly named, you’re a risk it routes around.

The second is corroboration. Models lean on what the web says about you, not just what you say about yourself. Ahrefs’ analysis found brand mentions correlate roughly 3× more strongly with AI visibility than backlinks, and that brands in the top quarter by web mentions get about 10× the AI visibility of the bottom three-quarters. Being talked about, accurately and by credible sources, is now a primary signal.

The third is quotable substance. An engine needs something concrete to lift. Pages dense with real statistics, named sources and direct answers get pulled into responses. The Princeton GEO study found such material lifts visibility by up to 40%. Notably, it does not have to be on a top-ranking page: only about 37% of AI Overview citations come from the top 10, down from roughly 76% a study earlier, so a precise statement on an unremarkable page can still be the source an answer is built from. Schema markup supports the identity piece, but it won’t earn citations by itself.

Why most brands are missing at least one

The reason so few names dominate is that each lever is necessary and none is sufficient alone. A pristine entity attached to thin content gets ignored. Brilliant content under an ambiguous identity never gets attributed. Strong third-party mentions mean little if the model can’t resolve who you are. Businesses tend to invest in one, usually content and sometimes schema, and assume the citations will follow. They don’t, because the system only works when all three line up and stay lined up as engines change.

That’s also why citation is defensible once you have it: it’s hard to assemble by accident. The brands that get named did the deliberate work of becoming identifiable, corroborated and quotable at once. You can see the pattern in practice in how a Hong Kong law firm became Perplexity’s first-named source, and the underlying ideas in what GEO is and what AEO is.

Engineering all three, and holding them as engines quietly change how they retrieve, is sustained specialist work. If you’d rather be one of the names the AI keeps returning to, that’s the work I do.

Frequently asked questions

Why does an AI cite some companies and not others? Because it synthesises from signals, not at random. It favours brands it can identify unambiguously, that credible sources corroborate, and that supply quotable, accurate substance. Miss any one and you’re easy to leave out.

Are backlinks still what matter most? Not for AI visibility. Brand mentions correlate roughly 3× more strongly than backlinks, and the top quarter by mentions gets about 10× the visibility of the rest.

Can I just add schema to get cited? No. Schema helps machines understand your content but isn’t a citation lever by itself. It supports a clear entity, it doesn’t replace authority and substance.