"How should we fix our website to do GEO well?" Cho Kyung-sang, the CEO of NNT, hears this question every week. Nine out of ten companies that come to him for consulting, he says, open their first meeting with it. And that approach he flatly calls "wrong." Not a different way of going about it—the wrong way.

GEO, or Generative Engine OptimizationGenerative Engine Optimization, has become an unavoidable term in 2026 marketing-budget discussions. As ChatGPT's search feature, Perplexity, and Google's AI Overviews begin to reshape how people look for information, companies are shifting part of their SEO budgets in this direction. The shift itself is a natural progression. The problem is that it is being built on the wrong frame. There is a clear reason why fixing your website does not make you show up better in AI search.

Why What Worked on Crawlers Doesn't Work on AI

SEO is built on how Google's crawler reads a web page. Meta tags, internal link structure, page-load speed, schema markup—these elements raise the Google bot's comprehension, and that in turn affects search rankings. The logic the SEO industry has accumulated over nearly two decades is clear: structure your site so the crawler can read it well.

Ask GEO the same question and it falls apart, for a reason. The way AI search works is fundamentally different. When Perplexity is asked, "Recommend a domestic marketing-automation tool," it searches the web in real time while simultaneously synthesizing learned patterns and external signals. In this process, the criterion the AI uses to choose its sources is not "how accurate are this page's meta tags." It is "how trustworthy are this brand, this author, this source on this topic."

The AI is not reading the technical structure of a particular website; it is looking at how that brand or person is cited and mentioned across the internet as a whole. Has the name appeared in trade publications? Do other experts mention it? Are there people who cited that source when answering questions in the field?—traces like these connect more directly to the AI's judgment.

If SEO is the work of announcing "I'm over here," GEO is the work of convincing others that "I know this field." The party being convinced has changed from the crawler to the AI. The crawler operates by rules, but the AI chooses on the basis of authority. That difference changes the entire approach.

Fixing your website is not entirely pointless. Because AI search does draw on web crawling to some degree, basic technical optimization is still necessary. The point is simply that it should not be the starting point of a GEO strategy. Before you spend millions of won (several thousand dollars) on a website overhaul, there is something you should ask first: "Is there any reason for AI to see our brand as an authoritative source?"

Why You Can't Be Unreservedly Optimistic About AI Search

Here an honest question has to be raised: whether there is, in fact, enough reason to focus on GEO right now.

The skeptical view is hardly trivial. AI search delivers answers, but in many cases it does not generate clicks. Even when Perplexity or ChatGPT names a particular brand inside an answer, the share of users who actually visit that site is, by some analyses, markedly lower than the rate of clicking on a Google search result. The SEO path in which exposure translated directly into traffic does not carry over intact to GEO.

Measurement is hard, too. There is not yet a standardized tool—something like Google Search Console—that lets you see in numbers how often your brand is referenced in AI search. Judging that "GEO is going well" is highly subjective. Some marketers even dismiss GEO as "content marketing dressed up under a new name." If consistently producing good content works for AI, for Google, and for people alike, do you really need a separate frame called GEO? That criticism carries weight.

Even so, the numbers are trending too clearly in one direction to ignore this channel entirely. Perplexity's monthly queries surpassed hundreds of millions as of 2024, and since ChatGPT added search, companies have been filing internal reports of shifts in traditional information-seeking patterns. Surveys also point to a rising share of B2B buyers who turn to AI tools when they first research a vendor. It is hard to deny that AI search has begun to enter the mix as one of the routes to being found.

What needs to be made clear at this point is that GEO does not replace SEO. The two are channels to run in parallel, and their priority and weighting vary by industry and target customer. For a small consumer-facing business, SEO on Google and Naver (Korea's dominant search engine) can still deliver more direct results. GEO is closer to long-term positioning. It is not a channel from which to expect short-term conversion.

What a Solo Operator Can Do Right Now

The fact that GEO has not yet fully settled makes now, if anything, a less crowded moment. Large companies have begun setting aside GEO budgets, but their execution is still tangled in confusion. To say that 90% are approaching it the wrong way is also to say that getting the direction right now means far less competition.

The first thing to check is your presence in outside media. AI treats content carried in trade publications, industry newsletters, and community contributions as far more authoritative than your own blog. Rather than posting more often on your own site, contributing articles to outside outlets, appearing in interviews, or making meaningful contributions in industry communities is more efficient from a GEO standpoint.

Creating information that exists nowhere else matters too. The content AI references tends to be the kind that contains concrete figures, original observations, or first-hand cases. Recording something you measured or experienced yourself—rather than a piece that summarizes what someone has already written—creates a reason to be cited. The on-the-ground experience a solo operator holds is a scarce ingredient in the AI-search environment.

Narrowing your subject area is worth considering as well. AI tends to recognize a source that has continuously produced content on a particular topic as a voice in that field. A deep, steady accumulation of traces in one area is more advantageous than covering many topics broadly.

One practical check is worth doing. Search your own name, your brand name, or the field in which you consider yourself an expert on Perplexity or ChatGPT. If the AI doesn't know you, that is a signal that you have left no trace worth citing within today's internet ecosystem. Deciding your next move after seeing that result is more productive than pulling out a website-overhaul proposal first.

Despite the impression the phrase "AI search marketing" gives, the starting point of this work is not technology. A single external trace worth an AI's reference comes before a single line of website code. Simply reversing that order puts you in a different position from most of the competitors standing at the same starting line.