The Visibility Problem Nobody Is Talking About
Whether you’re a longtime digital marketing leader or just someone trying to get some eyes on their product or company, you’ve probably heard a lot about GEO (or AIEO, AEO, or one of the other many acronyms) in recent months. Everyone is vying for a way to get tools like ChatGPT, Gemini, Perplexity, and Claude to understand what they do and, ideally, recommend them when a user asks a relevant question. But while most of the conversation around GEO is focused on content strategy and keyword architecture, many companies are completely overlooking a massive problem lying right in front of them: their own websites.
The visibility problem starts at the crawl, where search programs hunt for webpages to catalog and draw information from. Companies like Google don’t exactly parade this kind of information around, but these kinds of content-scraping tools only really see the first 15MB of any given webpage, and for AI tools it’s believed to be even smaller. That means that if a crawl doesn’t get to the meat of your company’s value proposition, product, or pitch, no AI system is going to even know it exists.
For a significant portion of enterprise websites, the most popular AI tools are working with a woefully incomplete picture. Your site might look great to a human visitor, but if it’s not built with an AI crawl in mind, it might be working against you.
What an AI Engine Actually Sees
When an AI engine encounters a website it processes the raw data it can see. Things like text, headings, entities, and internal links that tell it about topical relationships across the site. It evaluates content clarity, consistency, and structural logic, but it isn’t experiencing the site in the same way as a human visitor.
The things that make a website appealing to a living, breathing internet user carry no weight with an AI crawler. It doesn’t care if you have a flashy carousel or motion elements or interactive widgets unless it’s all supported by a clean and easily digestible data structure underneath. A site can be visually sophisticated or downright flashy, while still being entirely opaque to the systems that will ultimately determine how an AI tool references or recommends it.
It’s this massive gap between how a site looks and how an AI reads it that is often very substantial. The unfortunate truth is that most marketing teams don’t know this gap exists at all.
Where JavaScript-Heavy Sites Break Down
JavaScript frameworks are a fundamental piece of the modern internet, and they work quite well for user-facing purposes. The problem for AI systems attempting to index a site that is heavy on JavaScript is that it introduces latency and can’t always be completely rendered at scale.
If a tool like Gemini lands on a JavaScript-heavy site, it may only retrieve a partially rendered page, a bare bones skeleton of the full content, or in some cases nothing at all. For sites where things like descriptions of products, service breakdowns, and even blog and thought leadership content live within dynamic components, the content might never be indexed. To a human user, it’s clear and obvious, but from the perspective of the AI engine it may as well not exist at all.
Furthermore, use of high-res imagery, autoplaying video, and other asset-heavy design choices inevitably increase page load times and, as a result, reduce crawl efficiency. When less high-value content is properly indexed, the content signals in the AI systems are weaker, reducing the visibility of the material within generative search results. The unfortunate truth is that all too often it’s the technical decisions that create the most beautiful user experience that also degrade AI readability.
The Most Common Mistake: Prioritizing Features Over Foundations
GEO is still incredibly new and organizations are still grappling with what makes the most sense when it comes to website preparation. The most frequent missteps being made right now are when organizations treat GEO as a content problem, when at its core it’s a structural one.
Companies will invest in AI-generated content, roll out advanced chatbot tools, or invest in high-tech conversational interfaces while leaving the underlying architecture completely untouched. Adding these features on top of a site that wasn’t built for GEO does nothing to compensate for poor crawlability, inconsistent page structure, or internal linking that fails to reinforce messaging.
Hidden content can be detrimental, and the aforementioned JavaScript-based UI elements like tabs, accordions, and expandable sections that aren’t accessible within the initial HTML are essentially invisible to AI engines. The same applies to pages with thin or duplicate content, inconsistent heading hierarchies, and navigation structures that make it difficult for a crawler to understand the site’s topical scope.
Put simply: AI-friendliness is not something that can be layered on top of a site that is already poorly laid out. Accessibility and consistency at the architectural level is key, and should be prioritized ahead of added features and supplementary strategies.
Building for AI Readability: Where to Start
If you’re part of a team that’s in the midst of building (or rebuilding) a website with GEO in mind, there are three priorities worth addressing upfront:
The first is semantic HTML with a clear and consistent heading structure. Using clean, well-organized HTML gives AI engines a reliable map of each page. Schema markup adds context about entities, products, people, services, and events that give the AI a more accurate impression of what you’re offering. These aren’t technically complex considerations, but they are hugely impactful in how AI engines understand your website.
Next, focus on your internal linking architecture. A logical internal linking approach reinforces topical relationships across the entire site and goes a long way toward helping AI systems grasp the depth and breadth of a company’s expertise. Pages should always link to related content in a straightforward way, emphasizing subject-matter relationships rather than merely acting as a navigational aid. Sites that excel at this give AI engines a clear picture of what the organization knows and offers.
The third area of focus should be ensuring that content written in natural language that directly addresses user intent. Generative AI systems are trained to surface content that answers questions clearly and completely, and keyword-heavy content that is thin on substance is penalized. Writing that gets to the point, uses natural phrasing, and addresses real questions is more likely to be retrieved and cited by AI systems than content optimized for older search paradigms.
What Comes Next
GEO is still an emerging discipline, and we can be certain that best practices will continue to evolve as new models are introduced and optimizations are made. But the direction of travel is clear: Search is becoming generative. Users are increasingly receiving direct answers rather than ranked lists of links. The organizations that will perform best in that environment are the ones that make their content genuinely accessible and legible to AI systems, not just optimized for them on the surface.
Entity-based optimization, content depth, and contextual relevance are the long-term variables that will determine visibility in generative search. But none of that work pays off if the structural and technical foundations are not in place first. For most organizations, the most important GEO investment they can make right now is making sure their website can actually be read.
Billy Wright is Head of AI at Direct Online Marketing (DOM), a GEO-first digital marketing agency.
