Tech

React SEO: Why Client-Side Rendered Pages Lose Visibility in Google and AI Search

Client-side rendered React pages lose visibility for two separate reasons. Google can render them, but common implementation mistakes still hide content, links, or status codes from it. AI crawlers are a blunter problem, because most of them never run the JavaScript at all. Knowing which one you’re facing decides whether you need a small code fix or a different rendering setup.

What a React App Sends Before JavaScript Runs

A client-side rendered React page usually starts as nearly empty HTML, often little more than <div id=”root”></div> and a script tag. React’s own documentation notes that visitors see a blank page until the JavaScript loads and runs. Bots start from that same response, and everything after it depends on who’s reading.

Googlebot queues pages with a 200 status for rendering, then runs them in headless Chromium. A Vercel and MERJ study of over 100,000 Googlebot fetches on nextjs.org found that its 200-status HTML pages were fully rendered, so the claim that Google can’t read React is outdated. Rendering still isn’t guaranteed. Pages with noindex in the initial HTML weren’t rendered at all.

AI crawlers are a different story. A separate Vercel analysis from December 2024 found that none of the major AI crawlers it tested rendered JavaScript. ChatGPT’s and Claude’s crawlers did download script files, but they didn’t execute them. If a product description only exists after hydration or an API call, those bots have nothing to read.

Five React Mistakes That Cost Google Visibility

Most Google problems in React come from implementation choices, and rendering is only part of it. These five cause the most trouble, and each has a direct fix:

Mistake What happens Fix
Hash routes such as /#/products Google can’t rely on fragments to load different content Use History API routing
Links built with onClick Google may not extract the URL Render an a element with an href
Missing products return 200 Client-side routes create soft 404s Redirect to a real 404 URL, or add noindex with JavaScript
noindex removed by a script Google skips rendering, so the script never runs Keep noindex out of the initial HTML on pages you want indexed
API or JS files blocked in robots.txt Google won’t render blocked files, and content goes missing Allow the files that the page needs

The link problem is especially common because it looks fine in a browser, where a button that calls navigate(‘/shoes’) works for a person but gives a crawler no URL to follow. React Router’s Link component avoids this. It renders a real anchor tag.

Titles need less worry than people expect. Google accepts titles and meta descriptions set through JavaScript, and React can render a title from any component and place it in the document head. That helps Google. A bot reading only the raw HTML still won’t see it on a purely client-side page.

How to Check Your Own App

You can test this in a few minutes on one URL from each template. Run all three checks:

  1. Open view-source: and look for the main text, title, canonical, and internal links.
  2. Run URL Inspection in Search Console, then read the rendered HTML from the live test.
  3. Search for a sentence that only appears after rendering, using a site: query.

If the text is missing in step 1 but present in step 2, Google can probably handle the page while most AI crawlers can’t. If it’s missing from both, you’ve got a rendering bug to fix first, and no choice of architecture will help until that’s solved.

Choosing a Fix: SSR, Static Rendering, or a Rendering Layer

The right fix depends on how often your content changes and how much of the app you can afford to rebuild. Google recommends server-side rendering, static rendering, or hydration, and it describes dynamic rendering as a workaround.

Approach First HTML has content Best for Trade-off
Client-side only No Logged-in apps and dashboards Public pages depend on rendering
Server-side rendering Yes Pages with frequently changing data Needs a server, and hydration must match
Static rendering Yes Docs, blogs, and stable landing pages Content updates need a rebuild or revalidation
Rendering layer for bots Yes, for bots Apps that can’t migrate soon Adds a dependency, and Google calls it a workaround

In React itself, renderToPipeableStream sends HTML from the server, and hydrateRoot attaches React to it in the browser. Most teams won’t wire that up by hand. Next.js handles server and static rendering by route, and it can generate page metadata through its metadata API, which also solves the title problem for bots reading raw HTML.

Migration isn’t always realistic, especially for a large single-page app with a busy roadmap.

For React SEO on those apps, EdgeComet renders pages for bots in its own headless Chrome pool and serves cached copies in under 15 ms, while people keep getting the existing client-side app. A render missing the sections you’ve marked as required stays out of the cache. The feature starts on the Business plan at $199/month. Bots must get the same content users see, or this turns into a cloaking risk.

Fix the Cheap Problems First

Start with the table of mistakes, because hash routes, broken links, and soft 404s hurt under any rendering setup and rarely take long to fix. Then decide on architecture based on how much your traffic depends on public, frequently crawled pages, and keep logged-in screens client-side, since no bot needs them. Moving to SSR won’t save a page that still links through onClick handlers.

Related Articles

Back to top button