Your Engineers Built a Beautiful Website That AI Cannot See

Google can read your site just fine. ChatGPT, Perplexity, and Claude see an empty page.

There’s a test that takes ten seconds and might change how you think about your entire web stack.

Open your website in Chrome. Go to Settings → Developer Tools → press Ctrl+Shift+P → type “Disable JavaScript” → reload the page.

If your product descriptions, comparison tables, pricing details, and technical content all disappear — replaced by an empty shell with a spinning loader — then every AI crawler on the market sees the same empty shell when it visits your site. Your content doesn’t exist in ChatGPT’s world, Perplexity’s world, or Claude’s world. It only exists for Google, which is the one crawler that bothers to run your JavaScript.

That’s not a hypothetical. Vercel and MERJ tracked over 500 million GPTBot fetches and found zero evidence of JavaScript execution. Not partial. Not delayed. Zero. GPTBot downloads JavaScript files 11.5% of the time — it treats them as text data for training on code. It never runs them to build the page.

The same is true for every other AI crawler. ClaudeBot downloads JavaScript files in 23.84% of its requests — and never executes them. PerplexityBot doesn’t render JavaScript. Neither does Meta’s ExternalAgent or ByteDance’s Bytespider. Googlebot is the only major crawler with full JavaScript rendering support, using headless Chrome in a two-phase indexing system that Google spent over a decade building.

This creates what I call the Rendering Gap: the structural divide between what Google can see and what every AI system sees. And it’s invisible in your analytics.

Photo by frank mckenna on Unsplash

Why your dashboards don’t warn you

Your Google Search Console shows healthy crawling. Your organic traffic is steady. Your SEO team reports green across the board. Everything looks fine.

It looks fine because Google renders your JavaScript. Google indexes the content. Google ranks the pages. Every metric you track runs through Google’s infrastructure, which is the one system in the world that has invested in processing JavaScript-heavy websites.

But ChatGPT doesn’t search Google. It searches Bing — and 92% of ChatGPT’s citations match Bing’s index. Bing has partial JavaScript support at best. Perplexity maintains its own index. Claude searches Brave. None of them render your JavaScript.

The disconnect is measurable. 80% of URLs cited by ChatGPT, Perplexity, and Copilot don’t rank in Google’s top 100 for the original query. Your Google rankings and your AI citation reality are measuring different universes — and your dashboards only show you one of them.

A page that ranks on Google and converts visitors can be completely invisible to the AI engines that are now driving a growing share of B2B buyer research. Those two facts coexist silently — which is why this problem compounds before most teams notice it.

The scale of the problem

This isn’t a niche concern for a handful of single-page applications. Modern enterprise web development has moved heavily toward JavaScript frameworks. React, Vue, Angular, Next.js (when misconfigured for client-side rendering) — these are the standard tools enterprise engineering teams use to build websites in 2026.

GPTBot processes roughly 569 million requests monthly. ClaudeBot handles about 370 million fetches per month. PerplexityBot recorded a 157,490% increase in raw requests between May 2024 and May 2025 — the fastest-growing crawler on the web. AI crawlers now account for approximately 28% of total crawl traffic to major websites.

Over a billion AI crawler visits per month. All of them seeing raw HTML. For any site relying on client-side rendering, all of them seeing nothing.

The business impact is no longer theoretical. ChatGPT alone has 810 million daily users, and accounts for 77.97% of all AI-driven referral traffic to websites. AI referral traffic is growing roughly 1% month over month across all industries — Adobe reported a 693% surge in AI referral traffic during the 2025 holiday season alone. Google AI Overviews now appear in 25% of all Google searches, up from 13% in March 2025. This isn’t a niche channel. It’s becoming the primary discovery layer for B2B buyer research — and your JavaScript-rendered pages don’t exist in it.

When we run GEO audits for B2B companies, the JS rendering issue is one of the first things we check — and one of the most common findings. Marketing teams that have invested heavily in content quality discover that their engineering team’s architecture decision, made years ago for perfectly valid UX reasons, has made their content invisible to the fastest-growing discovery channel in B2B.

The content team and the engineering team are optimizing for different systems without knowing it. The content team builds pages that perform well in Google. The engineering team builds infrastructure that renders beautifully in browsers. Neither team has tested whether the content exists when JavaScript is disabled — because until 2024, no one needed to.

The Rendering Gap costs more than you think

In a previous article, I shared data from 8 GEO audits across B2B companies. Third-party domains owned 82–98% of all AI citations for our clients’ buyer queries. The brand’s own website accounted for just 2–18% of citations.

Some of that gap is content architecture — product pages don’t get cited the way buyer guides do. But some of it is the Rendering Gap. If your content literally doesn’t exist for AI crawlers, it can’t be cited regardless of how well it’s written. You could build the world’s most citation-worthy buyer guide, and if it renders via client-side JavaScript, AI crawlers see <div id="root"></div> and move on.

The cost isn’t just lost citations. It’s lost competitive intelligence. When your content is invisible to AI, you don’t show up in citation audits at all — which means you can’t even diagnose the problem using standard GEO methodology. You look like you have a content problem when you actually have an infrastructure problem.

Photo by Alex Litvin on Unsplash

The fix is known. The coordination is hard.

The technical solution is well-established: server-side rendering (SSR). Frameworks like Next.js, Nuxt, and Angular Universal support it natively. When your server renders the HTML before sending it to the browser, both human visitors and AI crawlers see the full content. The page works the same for everyone.

Pre-rendering is the alternative if full SSR migration isn’t feasible — generating static HTML snapshots of your pages that get served to crawlers while browsers get the JavaScript-powered version.

The technology isn’t the hard part. The coordination is.

The results, however, are fast. One company documented their journey from zero AI visibility to citations across all major platforms: Perplexity showed citations within two to three weeks of publishing server-rendered, structured content. ChatGPT followed at about four weeks — slower because it depends on Bing’s reindexing cycle. In a separate study, sites using SSR that loaded in under 2.5 seconds received up to 3× more AI citations than comparable client-rendered sites. Ahrefs found that sites with strong internal linking and silo structures earn 34% more AI Overview citations — structure that only works when AI can actually see it.

The fix doesn’t just open the AI channel. It often improves traditional search too. In one documented case, Google only rendered 20 of 59 JavaScript-dependent product pages. After SSR implementation, visibility improved across both AI crawlers and Google itself.

But SSR migration touches the engineering team’s architecture, the marketing team’s content pipeline, the DevOps team’s deployment infrastructure, and potentially the legal team’s consent management (cookie banners often rely on JavaScript too). In enterprise organizations, this is a cross-functional project that competes with product roadmap priorities.

This is the same cross-functional coordination problem we identified in our research on where GEO lives inside the enterprise. Marketing builds the content. Engineering manages the CMS. Legal controls the consent platform. Each function optimizes for its own objectives. Nobody optimizes for AI visibility — because nobody owns it.

The ten-minute diagnostic

Before starting any SSR migration conversation, run this audit:

Step 1: Disable JavaScript in your browser and load your ten most important pages. If the content disappears, you have the Rendering Gap.

Step 2: Check your robots.txt. Do you allow GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot, and anthropic-ai? If any are blocked — intentionally or accidentally — that’s a second invisible wall.

Step 3: Submit your sitemap to Bing Webmaster Tools if you haven’t already. ChatGPT searches Bing, not Google. If Bing hasn’t indexed your content, ChatGPT can’t find it.

Step 4: Check whether Brave has indexed your key pages. Claude searches Brave. Most companies have never checked.

These four steps take under ten minutes and cost nothing. They’ll tell you whether you have a content problem or an infrastructure problem — and the fix path is completely different for each.

In one documented case, Google only rendered 20 of 59 JavaScript-dependent product pages. After SSR implementation, visibility improved dramatically — not just for AI crawlers, but for Google itself. The rendering fix doesn’t just open the AI channel. It often improves the traditional search channel too.

The uncomfortable conversation

If you’re a marketing leader reading this, you now need to have a conversation with your engineering team. Not about content strategy or keyword optimization — about rendering architecture. The conversation goes something like this:

“When AI crawlers visit our site, do they see our content? Or do they see an empty div?”

Most engineering teams haven’t been asked this question. They built the site for browsers and for Google. Both work. But “works for Google” and “works for AI” are no longer the same statement — and the gap between them is growing wider every month as AI’s share of buyer research increases.

The companies that close the Rendering Gap now do it when the fix is a technical project. The companies that wait will close it when it’s a competitive emergency.

And the gap compounds. Once a brand becomes established as an authoritative AI source, it gets cited more frequently — creating a virtuous cycle where early visibility begets more visibility. The brands that fixed their rendering architecture six months ago are already accumulating citation authority that later movers will struggle to displace. Every month your content stays invisible is a month your competitors’ content gets cited instead — and the AI learns to prefer them.

I’m Sebastian Mueller, Founding Partner at MING Labs and founder of Hyperize. Over the past six months, I’ve published a 15-article series on AI visibility for B2B companies covering everything from the audit methodology to the content architecture that wins citations, from Chinese AI platforms to honest corrections where the data proved me wrong. If you want to know what AI tells your buyers about you — and who it recommends instead — that’s what we do.