Who Actually Wins in AI Search: Established Brands or Challengers?

The role of a website in the customer journey is changing. More discovery and consideration can now happen directly within search results and AI platforms, before a user ever reaches a brand’s website — if they visit it at all.

As more of the journey happens within these search experiences, the visibility brands earn there becomes more valuable. In AI search, the brands an LLM mentions, recommends and cites can shape consideration without requiring a click to their website. That’s made generative engine optimization increasingly important for brands looking to influence how and where they appear in AI search.

It also raises a question the search industry hasn’t settled: who actually earns that visibility in AI answers?

Some argue the established, long-standing brands that have come to dominate organic SERP results over the past few decades have the best shot at winning in AI search. But there’s a case to be made for nascent brands with little historical search authority and an eye toward breaking into AI search as the playing field gets leveled.

Whether a brand shows up depends largely on how the model builds its answer: by searching the web in real time, or by drawing on what it already remembers. Smaller brands have a shot at the first, while long-standing brands dominate the second. And a strong organic ranking doesn’t guarantee a place in either.

What is zero-click search?

Zero-click search is when a search query gets answered directly within search results without a click to a website, and it has been a growing trend for years. 

According to a June 8, 2026 SparkToro study, less than ⅓ of Google searches result in a click to a website. That number has been trending downward for years as search platforms look to keep traffic and users within their own ecosystems. This has only intensified with the rise of AI platforms and experiences, like Google AI Mode, which allows for simpler navigation within the platform and advanced information synthesis without clicking out of Google.

Resetting the marketplace

LLM responses are far more variable than traditional search results. Traditional search results vary by factors like location, but AI responses vary far more, often changing every time a prompt is run.

Research backs this up. In a 2026 preprint study called “Don’t Measure Once: Measuring Visibility in AI Search,” Schulte et al. tracked the same prompts daily across ChatGPT, Gemini, Google AI Mode and Perplexity over roughly 45 days. From one day to the next, the sets of cited sources overlapped by only 34–42% on average, meaning roughly two-thirds of the sources changed.

Running the same prompt repeatedly on the same day produced nearly the same instability. That suggests the shifts come largely from the models themselves, not just from updates over time. Brand mentions were somewhat steadier, overlapping 45–59% day to day, but still far from stable. That makes one-time measurement unreliable, and it means a single snapshot can overstate or understate how visible a brand really is.

There’s also a lot of variety between the models themselves. Gemini, Claude, ChatGPT, Perplexity and Copilot don’t all produce answers the same way. That’s why your brand needs a multi-model AI visibility strategy. While LLMs draw on some traditional SEO ranking factors to decide which content will help a user, they also weigh factors a traditional SEO strategy often doesn’t address.

Smaller brands are finding an opening

This variance can level the playing field for smaller publishers without a strong history of ranking in traditional search. It creates an opening in AI answers that organic search hasn’t had in years.

That’s evidenced well in this study from May 2026. Kirsten et al. in “Characterizing Web Search in the Age of Generative AI” find that 53% of the domains that Google AI Overviews is citing don’t show up in that same query’s top 10 blue link organic results. On top of that, 27% of domains in those AI Overview responses aren’t even in the top 100 search results of a traditional Google search.

This tells us 1 out of every 4 cited domains in Google’s AI Overview is a functionally non-existent domain in traditional search for that query. If that isn’t striking enough, basically half of those citations in AI Overviews are domains that don’t appear on the first page. While a legacy player might make an appearance in the AI Overview and also in the top-3 search results, half of the time, this new, premier AI Overview space is being given to a domain that most would never see otherwise. That’s an incredibly meaningful development within the search industry, and a huge opportunity for smaller brands looking to break into AI search results.

Larger brands are holding their ground

There’s also the case for AI search helping the established, larger players in the marketplace. 

Large brands with established rankings also have advantages in the new LLM landscape. In many cases, good SEO is also good GEO. AI models tend to rely on search engines to help them find information, in addition to their memory retrieval based on data from their model training. If you can be found easily in a search engine, there’s a good chance you have an opportunity to be mentioned or cited in an LLM response.

Larger brands also tend to have high authority signals and a detailed web of backlinking that traces back over years, even decades, which helps provide citation authority for LLM responses. Having the resources to provide website content in easy-to-read-and-retrieve formats is also good SEO that is good GEO. This foundational work will continue to help those who have been doing it for a long time.

How the model finds its answer matters

One thing that’s becoming increasingly evident is that an AI engine tends to do one of two things: search the web for a response or answer from its training data. Answers from training data are limited to what the model previously learned, which can create an echo chamber of the brands it already knows and leave new players out.

Sharma’s study “The Discovery Gap” (preprint, not yet peer-reviewed) illustrates this. It tested 112 startups from Product Hunt’s 2025 leaderboard across 2,240 queries to two models: ChatGPT (gpt-4o-mini, answering from training data alone) and Perplexity (with live web search). When asked about a startup by name, ChatGPT addressed it in 99.4% of responses.

But in open-ended discovery queries like “What are the best [category] tools launched in 2025?”, those startups surfaced only 3.32% of the time, a 30:1 gap. Perplexity, searching the web in real time, did better: an 8.29% discovery rate, with 27.7% of the startups appearing at least once, compared to 5.4% for ChatGPT. Its results also tracked familiar SEO signals, with referring domains the strongest predictor of whether a startup showed up.

So who actually wins?

The research illustrates two different sides of AI visibility: real-time retrieval can open the door to lesser-known brands, while memory-based discovery can reinforce the advantage of established ones.

Who actually wins?

AI search is not simply a reshuffling of the organic rankings, but a fundamental reshaping of visibility. In the narrow view of LLM prompting vs. traditional searching, the options for mention and citation are highly curated and wide ranging, opening up value for many different competitors. In the larger view, users can refine and redirect their prompts in real time, changing which brands get mentioned and cited with every follow-up.

Which brands benefit depends on how the model builds its answer. Fan-out queries, where the model breaks a prompt into several related web searches and pulls from the results, have the ability to deliver more specific results for a given user and uncover more resources than ever before, opening the marketplace up to a lesser-known brand with a more refined response to address a user prompt. Memory retrieval, on the other hand, tends to reinforce existing knowledge and the brands the model already knows.

Currently, there are no reliable signs of which prompts will trigger a memory retrieval vs. a fan-out search, and it can vary by model. Brands in some industries may find that most of their target audience uses the types of prompts that lead to memory retrieval. Other industries may not be so straightforward. The only way to know which applies to you is to track your visibility across the prompts your audience actually uses, which is exactly what we built PAVE, our AI intelligence engine, to do.

That’s what sets AI visibility apart from traditional search: a strong organic ranking helps, but it guarantees a place in neither type of answer.

What this means for your brand

Plenty of things have “killed” SEO before, and yet we’re still here seeing the benefits of it being done well. Zero-click search is no exception. While some may exclaim it has finally killed SEO, brands that are resilient and ready to adapt will succeed through it.

What can you do to come out on top?

If you’re a smaller or newer brand

Stay competitive. What that looks like depends on where you’re starting from. If you’re a smaller or newer brand, focus on the opening that real-time search creates. Create content that answers specific questions clearly enough to be cited, and build mentions beyond your own website, such as reviews, industry publications and partner pages, so your brand has a stronger presence across the sources AI systems may draw from.

If you’re an established brand

If you’re an established brand, don’t assume your rankings carry over. Your history is an advantage in memory-based answers, but it doesn’t protect you in real-time ones. Look at how often you’re actually mentioned and cited across AI platforms, and which competitors are showing up instead.

For all brands

For all brands, ensure verified data informs every strategy decision you make. Because AI visibility shifts from prompt to prompt and week to week, you need to measure more than once and track results over time and across platforms. As has always been the case in SEO, there are tactics that have worked for ages and some that come and go. Consistent measurement is how you’ll tell the difference.

Stick to your SEO strategy if it’s working. It’s already doing legwork for your AI visibility if done right. That being said, begin to think about what your goals are in AI search. Depending on what those goals are (visibility, referrals, leads, etc.), a tailored GEO strategy will complement your SEO work and set a solid foundation for the future with AI.

Show up when it matters with Perrill

Whether you’re an established brand protecting your ground or a newer one looking for an opening, Perrill can help you build an SEO and GEO strategy grounded in real data, not one-time snapshots. Connect with our team to see where AI search can take your brand.