You've probably heard a lot lately about optimising your website for AI Search. Answer Engine Optimisation (AEO), AI search, generative search, and LLMs (large language models) - the terminology is evolving fast! But most of the conversation focuses on what to write. Very little of it addresses a more fundamental question - can AI actually read what's on your website in the first place?
The answer, perhaps surprisingly, is - not always. Understanding why and what to do about it could give you a significant edge over competitors who are optimising their content without ever asking this question.
How AI Search Tools Actually "Read" A Webpage
When most AI systems fetch a webpage, it doesn't experience it the way you do in a browser. There's no rendering, no visual layout, no images loading, it simply receives the raw HTML code that the server sends back and reads that as text.
For traditionally built websites, this works well. The HTML contains all the content, the headings, paragraphs, and data, right there in the code, ready to be read. But here's where it gets interesting…
The JavaScript Problem
A huge proportion of modern websites are built using JavaScript frameworks. These are powerful tools that create fast, dynamic, visually impressive websites. But in the background, they work in a fundamentally different way. Instead of sending a page full of content, the server sends back a bare HTML shell, essentially just enough code to load the JavaScript framework. The actual content is then fetched from an API and assembled dynamically, live in the user's browser.
When a human visits the page, their browser runs the JavaScript, and they see the finished product. When an AI tool fetches the page, it can receive just the empty shell. Effectively the browser doesnt run and the content never assembles. The numbers, articles, and data that a human can clearly see simply don't exist yet at the point the AI tool is “reading”. You could think of it like being handed a flat-pack furniture box and being asked to describe the finished wardrobe. The components are all there, but without being able to assemble them, you can only see the packaging. What this means is that all this has very real implications for how AI systems understand and represent your website's content.
*However there is always an exception to the rule - Googles AI system Gemini can reportedly "see" a little more than some of the others. Since it is built on the Google platform it uses the same Web Rendering Service (WRS) that powers Google Search. This means when Gemini "looks" at a website, it is able to run the JavaScript to see the final, rendered page, similar to what we would see. But latency issues, infinite scrolling and Shadow DOMS still pose significant visibility issues even for Gemini.
What Else Can’t AI “See”?
JavaScript rendering is the most widespread issue, but it's far from the only visibility limitation AI systems face. Here's a fuller picture of what can also be “invisible”.
- Dynamic And interactive Content: Anything that only appears after a user interaction (clicking a button, expanding an accordion menu, typing into a search bar) is typically invisible to AI. The AI tool sees the page as it arrives, not as it behaves after interaction. Infinite scroll or Load More buttons are common examples; AI only sees the content present in the initial HTML response, not the items that load as a user scrolls down or clicks the load more button.
- Authentication-Protected Content: Anything behind a login wall is inaccessible to AI, including paywalled articles, members-only resources, and private dashboards. AI systems can't “log in” or solve CAPTCHA’s, so this content is effectively off-limits.
- Iframes: Content loaded from an external source inside an iframe, embedded maps, third-party tools, and external forms are typically invisible to AI systems reading the parent page.
- Real-Time And Live Data: Content delivered via WebSockets, live data feeds, or real-time dashboards generally can't be read. AI sees a snapshot of the initial page load, not a live stream of data.
- Bot-Blocked Pages: Many large platforms like LinkedIn, Facebook, and others actively detect and block automated fetching, returning either nothing or a very stripped-down version of the page. If your content lives primarily on these platforms rather than your own website, it may be far less visible to AI than you think!
- Video Content: This is a big one with significant marketing implications so lets dive in with a little more detail. Simply put, AI systems cannot immediately “see” what is depicted in a video. They cannot watch a video or listen to audio, but they can see that it exists and read any associated text in the HTML, but the visual content itself is invisible. When AI encounters a video on a webpage, I'm almost entirely reliant on surrounding text, transcripts, captions, titles, descriptions, and any other HTML content near the video. It cannot analyse motion sequences…yet… this would be an enormous computational task!
- What About Static images? AI tools do currently have genuine “vision extraction” capabilities for static images. For example, if you loaded an image directly into Gemini or ChatGPT, it can actually analyse and interpret what's in it with reasonable accuracy. What this means for your website in a nutshell is that fancy header video, dedicated explainer video, entire gallery of stunning location photographs, and infographic are making it harder for AI tools to “read” your content. You aren’t communicating anything useful to AI unless it is accompanied by descriptive text! This is actually one reason why video transcripts are so valuable, they are the primary way any AI system is going to understand video content for the foreseeable future.
Why AI Visibility Matters For Your Marketing
You might be thinking "This sounds like a technical problem for developers, not marketers." But the visibility of your content to AI has direct marketing consequences.
AI-powered search engines and answer engines are increasingly being used to answer the kinds of questions your potential customers are asking. When someone asks an AI "what are the best digital marketing agencies in Christchurch" or "how do I improve my website's SEO," AI tools draw on content they have been able to read and understand. If your best content is locked inside JavaScript rendering, hidden behind login walls, sitting in images without descriptive text or videos without transcripts, it simply doesn't exist from the AI's perspective.
This is the new frontier of Search Discoverability, and the businesses that get ahead of it now will have a meaningful advantage as AI-driven search continues to grow.
Why Content Length Matters Too
There's one more technical concept worth understanding when discussing this topic, and that is context windows and content depth. AI systems have a limit on how much text they can process at once, which is called the context window. For reference, a modern large language model might have a context window equivalent to around 150,000 words (roughly one and a half novels' worth of text). That sounds like a lot, and for most purposes it is. But when an AI is fetching multiple pages, processing long documents, and holding a conversation simultaneously, that budget gets used up pretty quickly.
The practical implication of this is that very long, long, long pages may get truncated. If your most important content, your key services, location pages, your strongest testimonials, or your clearest value proposition is buried at the bottom of a very long page, there's a real risk it gets cut off before the AI ever reaches it. This reinforces a principle that good content marketers already know and sometimes refer to as front loading, or in other words, lead with what matters most! But it also adds a new dimension for digital marketers - you're not just writing for humans who might lose interest and stop scrolling, you're also writing for AI systems that may literally stop reading.
8 AI Visibility Action Steps To Take Away
The good news is that the solutions to these issues are largely the same things that good web development, SEO and content production experts have always recommended. AI visibility and human accessibility go hand in hand.
1. Write Meaningful Alt Text For Every Image
Alt text is the single most impactful thing you can do for image visibility. A well-written alt attribute describes what's actually in the image in specific, useful terms. For example, alt="image1" is no good, but alt="Christchurch city centre at dusk with the Avon River in the foreground" is perfect! Not only does this make your images readable to AI, but it also improves accessibility for visually impaired users and gives search engines better signals for image indexing.
2. Use Descriptive Filenames For Images
Filenames are often overlooked as a potential optimisation technique. For example, waikato-river-hamilton-new-zealand.jpg communicates far more than IMG_4587.jpg to AI, to search engines, and to anyone managing your media library.
3. Always Always Always Provide Written Transcripts For Videos
If video is a significant part of your content strategy, transcripts are essential. A text transcript sitting on the same page as an embedded video means the entire content of that video becomes readable by AI systems, by search engines, and by users who prefer to read or who are in a noisy environment. It's one of the highest-value, lowest-effort content investments you can make.
4. Use Proper Semantic HTML
HTML was designed with meaning built in for good reason. Headings (<h1>, <h2>), paragraphs (<p>), lists (<ul>, <ol>), figures (<figure>, <figcaption>), these tags tell AI and search engines what kind of content they're looking at and how it's structured. A page built with semantic HTML is significantly easier to understand than one built entirely with generic <div> tags.
5. Implement Structured Data/Schema Markup
Schema.org markup is essentially a standardised vocabulary for describing your content to machines. You can markup articles, products, local businesses, FAQs, videos, events, and much more. When an AI or search engine reads your page, schema markup gives it precise, structured information about what your content is and means, removing ambiguity and improving how your content is understood and represented.
6. Consider Server-Side Rendering (SSR) For JavaScript Frameworks
If your website is built on a JavaScript framework, talk to your developer about server-side rendering. SSR means the server assembles the full HTML content before sending it, so AI systems and search engine crawlers receive a complete, content-rich page rather than an empty shell. Most modern frameworks support this, and it can make a substantial difference to both AI visibility and traditional SEO.
7. Keep Your Most Important Content In Plain HTML
For critical content, your core services, key value propositions, and contact information make sure it lives in plain HTML on the page, not locked inside JavaScript components, iframes, video or images. If it's important for humans to find via AI, it should be in textual content that AI can read easily first and foremost.
8. Structure Pages So Key Content Appears Early
Given the context window limitations discussed above, place your most important content near the top of the page (front loading). This is good UX practice anyway, but it also ensures that if an AI does truncate a very long page, it has already encountered your key messages before it stops reading.
The Bigger Picture - Accessibility, SEO, And AI Are Fast Converging
Perhaps the most important takeaway from all of this is that AI Search Visibility, traditional SEO, and web accessibility are not separate concerns requiring separate strategies. They are increasingly the same thing. Every improvement you make for AI readability, whether it's alt text, transcripts, semantic HTML, structured data, or server-side rendering, also improves your rankings in traditional search and generally makes your website cleaner, faster, and better structured.
The businesses that will win in the AI-driven digital landscape are not necessarily those who chase the latest optimisation trend. They are the ones who build their digital presence on solid fundamentals - content that is clear, well-structured, and genuinely accessible to humans and machines alike.
If you'd like to understand how visible your website is to AI systems and what practical steps would have the biggest impact for your specific situation, get in touch with the digital marketing team at Digital Influence today.
Interestingly, this article was produced with the assistance of several AI tools - which, as it turns out, gave us some rather useful first-hand insight into the topic!
