(Technical)

JavaScript SEO

Google reads the same page that AI can’t

Even if a page looks fine in a browser, what search engines and AI receive can differ. In some cases the text is already in the HTML the server first sends; in others the server sends only an empty shell, and JavaScript draws and fills in the page. To human eyes the result is the same, but to machines it is completely different.

JavaScript SEO is managing this difference. It refers to the work of making sites that render their pages with JavaScript readable by both search engines and AI.

Google runs JavaScript

Google processes JavaScript sites in three stages: crawling, rendering, and indexing. It fetches the page (crawling), runs the JavaScript in a browser-like environment to complete the page (rendering), and then adds the result to its search database (indexing). Google (official)

Rendering doesn’t happen immediately; pages wait in a queue. Because of this wait, the claim that “Google indexes JavaScript sites slowly” has circulated for a long time, but measurements show the delay is shorter than expected. From crawl to the end of rendering, half finished within 10 seconds, and three out of four finished within 26 seconds. MERJ/Vercel data

Coverage is broad too. In the same study, every HTML page examined was rendered, and content filled in by separate requests to the server after the page opened was all indexed. As far as Google alone is concerned, sites built with JavaScript have no major problems. MERJ/Vercel data

A Google Search Central video explaining JavaScript SEO. It appears in the same section of the document above.

Most AI bots don’t run JavaScript

This is where things split. The crawlers of ChatGPT and Claude do fetch JavaScript files: ChatGPT spends 11.50 percent and Claude 23.84 percent of all requests on JavaScript files. But they only download them; they don’t run them. To these bots, content that JavaScript draws in might as well not exist. Vercel data

A January 2026 study found the same. The bots of OpenAI, Anthropic, Meta, ByteDance, and Perplexity all failed to run JavaScript. Only three ran it: the bots of Gemini, Applebot, and Common Crawl. Gemini can do it because it uses Google’s crawling infrastructure as is. Search Engine Journal

Google’s official documentation points this out too. It recommends completing pages on the server before sending them, citing among its reasons that not all bots can run JavaScript. Google (official)

So if your site cares about GEO, you need to flip your thinking. Ranking well on Google is no guarantee of appearing in AI answers. Search visibility and AI citation have different gates to pass.

What decides it is the rendering method, not the tool

Using a tool like React or Vue doesn’t make a site unreadable, and building with WordPress doesn’t make it readable automatically. There is only one test: is the content inside the HTML the server first sends?

  • If the page is completed on the server (server-side rendering), AI bots can read it no matter what tool built it.
  • If the page is drawn in the browser (client-side rendering), Google can read it but most AI bots cannot.
  • Even on WordPress, if key content is filled in only by JavaScript, it is just as invisible.

It is true that WordPress tends to be relatively safe, but that is not because it is WordPress; it is because of its structure, which completes the HTML on the server before sending it. Look at the structure, not the name of the tool.

How to check which kind your site is

Open View Page Source in your browser to see the original HTML exactly as the server sent it. On Windows it’s Ctrl+U; on Mac it’s Command+Option+U. Search there for a snippet of your body text.

  • If you find it in the source, the content is in the static HTML. AI bots can read it.
  • If it isn’t in the source but shows on screen, JavaScript drew it. AI bots can’t see it.

The Elements tab in developer tools shows the final state after all JavaScript has run, so don’t use it for this check. Only comparing against View Page Source reveals the difference. Search Engine Journal

Content that search and AI will refer to, such as the title, description, body text, and menus, should already be present at the View Page Source stage. Vercel data

Search for a body sentence in View Page Source (Ctrl+U)
  Found     → static HTML  → both Google and AI can read it
  Not found → JS rendering → Google can read it, AI can't

Google doesn’t recommend dynamic rendering

Serving bots a separately generated, completed HTML page while showing people the original is called dynamic rendering. It was once used as a fix, but Google has concluded that it was only a workaround, not a long-term solution, because it makes the setup more complex and uses more server resources. Google (official)

Instead, Google recommends server-side rendering, static rendering, and hydration. Hydration means sending HTML completed on the server first, then adding JavaScript in the browser to make it interactive. Because the content is in the first response, bots can read it, and users keep the original interactivity. Google (official)

If you block JavaScript files in robots.txt, even Google can’t render the page, because it doesn’t run JavaScript from blocked files. Check this whenever you adjust your robots.txt settings. Google (official)

References

Aug.2026By Studio JT

More from Technical

Like this term?
Give it a like!

(38)