AI Crawl Radar
Fix guides 4 min read

Can AI Crawlers Read JavaScript? Fix JS-Only Pages

Can AI crawlers read JavaScript? Mostly not. See what GPTBot and ClaudeBot get from your page, and fix JS-only content, thin pages, H1s and alt text.

Cover image: Can AI Crawlers Read JavaScript? Fix JS-Only Pages
On this page
  1. What the check measures
  2. "Content appears to require JavaScript" (critical)
  3. "Thin content"
  4. "Add a single descriptive H1"
  5. Images without alt text
  6. Low text-to-HTML ratio
  7. Verify
  8. Related guides

Can AI crawlers read JavaScript? Mostly not. Most AI crawlers download your HTML and read it as it arrives. Unlike a browser, they generally do not run your JavaScript. If your page builds its text in the browser after load, those crawlers receive an empty shell, and your content might as well not exist for them. Googlebot is the exception: it renders JavaScript, so a page can rank in Google and still look empty to GPTBot, ClaudeBot or PerplexityBot. Google describes its own rendering in its JavaScript SEO basics.

Content Accessibility is worth 15 of the 100 points in an AI Crawl Radar audit. It checks whether a crawler that does not run JavaScript can still read your page, and whether what it reads is substantial and well structured.

Side by side: a browser renders a full page, while an AI crawler receives only an empty root div and a script tag
The same URL, as a browser and as an AI crawler see it.

What the check measures

SignalShare of the category
Main content is present in the initial HTML (no JavaScript needed)40% (5% if it seems to need JavaScript)
At least one H1 heading15%
Word count: 500+ words, or 250–49920%, or 10%
Fewer than 30% of images missing alt text (or no images)10%
Text-to-HTML ratio above 5%15% (5% otherwise)

The first row is the big one, and it is also where the most serious finding comes from.

"Content appears to require JavaScript" (critical)

The audit flags this when the raw HTML contains fewer than about 50 words of visible text and looks like a JavaScript application shell: a root element such as <div id="root">, id="app", id="__next" or id="__nuxt", or more than five script tags. That is the signature of a page that is rendered in the browser.

To see what a crawler sees, view the raw source of your page (View Source, not the Elements panel, which shows the page after JavaScript ran), or fetch it from a terminal:

curl -s https://yourdomain.com/ | sed -e 's/<[^>]*>//g' | head -50

If your headline and main paragraphs are not in that output, AI crawlers cannot read them either.

How to fix it

The fix is to send real HTML from the server. The report lists this as "Serve content server-side (SSR/SSG)". How you do it depends on your stack:

  • React: move to a framework that renders on the server, such as Next.js or Remix, or pre-render static routes at build time.
  • Next.js: make sure content pages use server components or static generation, and that the main text is not fetched in a client-side useEffect.
  • Vue: use Nuxt with server-side rendering or static generation.
  • Angular: use Angular's server-side rendering.
  • Single-page apps you cannot change: a pre-rendering service can serve a rendered snapshot to crawlers. Serve the same content to bots and users. Showing crawlers different content is cloaking, which search engines penalize.

Interactive parts can stay client-side. Only the content you want read and cited needs to be in the initial HTML: headings, body text, product details, FAQs. Widgets, charts and forms can still load with JavaScript.

"Thin content"

The audit counts the words in the visible text of your page. Under 250 words raises a Thin content warning; 500 or more earns the full word-count share. Answer engines quote pages that actually answer something, and a page of three sentences rarely does.

Fix: add substance, not filler. Explain what the page offers, who it is for, how it works and the questions people ask about it. Useful additions include specifications, step-by-step instructions, comparisons and a short FAQ. A product or landing page does not need 2,000 words, but it does need enough text to answer a real question on its own.

"Add a single descriptive H1"

The page has no <h1>. The H1 is the clearest signal of what a page is about, for search engines and for AI systems that extract passages.

Fix: give each page one H1 that names its topic in plain words, such as "Project management software for small agencies" rather than "Welcome". Use H2 and H3 headings below it to structure the rest. Many themes turn the logo into an H1 or use no H1 at all; check the HTML, not the visual design.

Images without alt text

If 30% or more of your images have no alt attribute, you lose this share. Alt text is what a crawler "sees" instead of the image.

Fix: describe what each meaningful image shows, as in alt="Dashboard showing weekly active users rising from 1,200 to 1,850". Purely decorative images should have an empty alt="", which tells crawlers to skip them.

Low text-to-HTML ratio

If visible text makes up 5% or less of your HTML's size, the page is mostly markup, inline scripts and styles. Crawlers download all of it to find a little text.

Fix: move large inline scripts and styles into external files, remove unused markup from your theme, and avoid shipping huge serialized data blobs in the page when you only need a fraction of them.

Verify

  1. Fetch the raw HTML with curl and confirm your H1 and body text are present.
  2. Re-run the AI visibility audit and check the Content Accessibility category.

Frequently asked questions

6 questions people ask about this topic

Can AI crawlers render JavaScript?
Mostly not. A Vercel and MERJ analysis of crawl logs found that OpenAI's and Anthropic's crawlers download JavaScript files but do not execute them. Googlebot does render JavaScript, so a page can rank in Google and still look empty to ChatGPT or Claude.
How can I see what an AI crawler sees on my page?
View the raw page source (not the browser's Elements panel), or fetch the page with curl. If your headline and main text are not in that HTML, crawlers that don't run JavaScript cannot read them.
Does my React or Vue site need server-side rendering?
If you want its content read by AI crawlers, yes, at least for the pages you want cited. Frameworks such as Next.js, Nuxt and Remix render HTML on the server; static generation works too. Interactive parts can stay client-side.
Is serving different HTML to AI bots cloaking?
Serving bots a rendered version of the same content is fine. Serving them different content from what users see is cloaking, which search engines penalize.
How many words does a page need?
The audit warns under 250 words and gives full credit from 500. Add substance that answers a real question, not filler.
Do images and alt text matter for AI crawlers?
Crawlers that don't run vision models only read the alt text. The audit gives credit when fewer than 30% of images lack alt text; decorative images should use an empty alt="".

See how AI crawlers read your site

Run a free AI visibility audit: a 0–100 score and a fix for every lost point.

Analyze my website

Keep reading

All guides →