JavaScript SEO Audit: Check That Google Sees Your Content
Short answer Open your page source and search it for a sentence from the page. If the text is there, JavaScript isn't hiding it. If not, run Search Console's live test and check the rendered HTML for your text, links, title, canonical and noindex. Most AI crawlers, including ChatGPT's and Claude's, don't run JavaScript, so key content belongs in the HTML.
Want help applying this to your business? Talk it through on a free 30-minute call.
Book a callDoes my website need a JavaScript SEO audit?
Only if your words or links appear after the page loads instead of arriving with it. Find out in two minutes, no tools needed:
- Open a service page in Chrome on a computer and press Ctrl+U (Cmd+Option+U on a Mac). That shows the raw HTML your server sends, the first thing a crawler gets.
- Press Ctrl+F and search for a sentence from the middle of the page, then for the address of a page your menu links to.
- Repeat on your home page and one blog post.
| What you find | What it means | Next step |
|---|---|---|
| Your text and links are all in the source | JavaScript isn't hiding anything | No JavaScript audit needed; any ranking issue lies elsewhere |
| Some text is missing (reviews, prices, FAQ, a menu) | Those parts are added by scripts | Audit those parts |
| The source is nearly empty, mostly script tags | The whole site is built in the browser (a "single-page app") | Run the full audit below |
Custom sites built in React, Vue or Angular are the usual "nearly empty" case. Template-builder sites can still load parts by script, such as review feeds, booking embeds or menus. Check rather than assume.
How does Google read a JavaScript page?
In three phases: crawl, render, index. Google's JavaScript SEO basics explain that Googlebot first fetches the HTML and collects its links, then queues the page so a headless Chromium browser can run the scripts, and finally indexes what that rendered version shows. Two details matter for an audit:
- Rendering waits in a queue. Google says a page may wait "a few seconds, but it can take longer than that". A 2024 study by Vercel and MERJ of over 37,000 renders on nextjs.org measured a median wait of 10 seconds and about 18 hours at the 99th percentile. Long waits are rare, but they happen.
- Only pages with a 200 (success) code are sure to be queued. Google sends every 200 page to rendering, but on an error page such as a 404, "rendering might be skipped".
Google also never clicks or scrolls. Its lazy-loading guide says "Google Search does not interact with your page." Anything that appears only after a click is invisible to it.
The audit: seven checks on your top pages
You need Search Console access (or Google's free Rich Results Test, which needs none) and 30 minutes. Do your five most important pages first: home, top services, contact.
| # | Check | How | Pass |
|---|---|---|---|
| 1 | Rendered text | Search Console, URL Inspection, Test live URL, View tested page, HTML tab | Every heading and paragraph you care about is in it |
| 2 | Screenshot | Same panel, Screenshot tab | The page looks complete, not blank or half-built |
| 3 | Errors | Same panel, More info: page resources and JavaScript console messages | No blocked scripts, no errors on files that build content |
| 4 | Links | Search the rendered HTML for your menu links | Each is a standard link tag with an href (address), not a button |
| 5 | Tags | Compare title, meta description, canonical, robots tags and structured data in source vs rendered | They match, and no noindex on a page you want found |
| 6 | Error pages | Open a made-up address, like yoursite.com/zzz-test | A "not found" page that returns code 404, or carries noindex |
| 7 | Hidden content | Look for tabs, "load more" buttons and endless scroll | The content is in the rendered HTML without a click |
Google's troubleshooting guide names these same tools: the URL Inspection tool or the Rich Results Test show "loaded resources, JavaScript console output and exceptions, rendered DOM". Search Console Help adds that the screenshot exists only in a live test, and that the default inspection view shows the last indexed version, "not the live version on the web".
For check 6, ask your developer or use a free "HTTP status checker" site. A missing page that returns 200 is a "soft 404": Google may index your error page as if it were real content.
What problems does the audit usually find?
| Problem | What Google says | Fix |
|---|---|---|
| Menu built from buttons, not links | Google can generally crawl only a standard link tag with an href address | Real links; scripts can still handle the click |
| Page addresses with `#/` (like site.com/#/services) | Googlebot "can't reliably resolve" them; use the History API instead | Give each page its own address |
| noindex in the HTML, removed later by script | Rendering may be skipped, so removal "may not work as expected" | Leave noindex out of the HTML on pages you want found |
| Canonical changed by script | Don't change it to a different URL from the one in the HTML | Set it once, in the HTML |
| Content loads on scroll or click | Load content when it is visible in the viewport; give each chunk of endless scroll its own URL, like ?page=2 | Ask your developer to follow Google's lazy-loading guide |
| robots.txt blocks script files | Google "won't render JavaScript from blocked files" | Allow your JavaScript and CSS folders |
| Old scripts after an update | Googlebot caches aggressively and may use outdated files | Fingerprinted filenames, like main.2bb85551.js |
The noindex row causes the most damage, and Google's wording is exact:
"When Google encounters the noindex tag, it may skip rendering and JavaScript execution, which means using JavaScript to change or remove the robots meta tag from noindex may not work as expected." (Google Search Central: Understand the JavaScript SEO basics)
Tabs and accordions are fine when their text is already in the rendered HTML, just folded away. Text fetched only on a click is the problem.
Do ChatGPT and other AI crawlers see JavaScript content?
Mostly no. Vercel and MERJ's December 2024 study of AI crawler traffic found that none of the major AI crawlers render JavaScript, naming OpenAI's GPTBot, OAI-SearchBot and ChatGPT-User, Anthropic's ClaudeBot, Meta's crawler and PerplexityBot. They fetch script files but don't run them. Gemini uses Googlebot's infrastructure, and Applebot renders pages like a browser. The study dates from late 2024 and crawlers change, but don't count on them running your scripts.
So a page that passes Google's live test can still look empty to an AI assistant. If you want to be recommended in AI answers, your services, areas and prices belong in the HTML itself. Where those assistants get local business facts is covered in where ChatGPT finds local businesses.
How do I fix JavaScript SEO problems?
Have the server send the finished HTML. Google's advice:
"Keep in mind that server-side or pre-rendering is still a great idea because it makes your website faster for users and crawlers, and not all bots can run JavaScript." (Google Search Central: Understand the JavaScript SEO basics)
| Approach | What it is | Google's view |
|---|---|---|
| Server-side rendering | The server builds each page's HTML on request | Recommended |
| Static rendering (pre-rendering) | Pages are built into HTML files ahead of time | Recommended |
| Hydration | HTML first, then scripts add interactivity | Recommended |
| Dynamic rendering | Bots get one HTML version, people get the script version | A retired workaround |
"Dynamic rendering was a workaround and not a long-term solution for problems with JavaScript-generated content in search engines." (Google Search Central: Dynamic rendering as a workaround)
Ask your developer which of these your framework supports; it may be a setting rather than a rebuild. After the fix, run the live test again on the same pages.
Which tools do I need?
| Tool | Cost | What it does | Limit |
|---|---|---|---|
| View source (Ctrl+U) | Free | Shows the HTML before scripts | Not what Google renders |
| Inspect (right-click, Inspect) | Free | Shows the page after scripts run in your browser | Your browser, not Googlebot |
| Search Console URL Inspection | Free | Google's own render, screenshot and errors | One URL at a time; needs site access |
| Rich Results Test | Free | Google's render for any public URL | One URL at a time |
| Screaming Frog SEO Spider | £199 a year for JavaScript rendering | Renders every page and compares | The free version doesn't include JavaScript rendering |
Screaming Frog's user guide lists JavaScript rendering among the features a licence unlocks. A 10-page site doesn't need it.
Example
Illustrative cases with invented numbers.
A law firm's 12-page site on a template builder
The owner finds every paragraph in the page source, but the Google reviews widget isn't there. The audit stops: nothing important depends on scripts. The reviews still show on the Business Profile.
A remodeler's custom React site
The source is nearly empty. The live test shows full text, but the menu items are buttons, and only 4 of 30 pages are indexed. The developer swaps the buttons for real links and turns on static rendering. A month later the Page indexing report shows 28 indexed.
FAQ
Can Google index JavaScript content?
Yes. Google renders pages in an up-to-date Chromium, but after a queue, without clicking, and maybe not at all on noindex or error pages.
Is JavaScript bad for SEO?
No. Scripts that add features to a page with real HTML cause no harm. The risk is content or links that exist only after scripts run.
Is "view page source" the same as what Google sees?
No. Source shows the HTML before scripts; Google indexes the rendered version. Compare both, and use URL Inspection for Google's own render.
How often should I run a JavaScript SEO audit?
After any redesign, platform move or framework upgrade, and whenever indexed pages fall in Search Console. Otherwise, live-test your top pages quarterly.
Different for you?
- Your pages aren't on Google at all: website not showing up on Google.
- Search Console marks pages as not indexed: Crawled or discovered, currently not indexed.
- Traffic fell right after a new site launched: traffic dropped after a redesign.
- You're choosing a website platform: is Wix or Squarespace bad for SEO?.
- Your script-heavy pages load slowly: Core Web Vitals and ranking.
- You want the site rebuilt to send real HTML: Website Agent.
- You want these checks run for you: SEO & GEO Agent.