Recent Posts
Tags Cloud
Lab 4: Is your main content in the raw page? The view-source test in 10 minutes
- November 19, 2026
- 0 Comments
- 0 Likes
Is your main content in the raw page, and do AI crawlers render JavaScript before they read it? Mostly they do not, so your main content is only safe when it sits in the raw HTML your server sends. This Lab gives you two free browser tests to find out in about 10 minutes.
By Elizabeta Kuzevska, Co-Founder, Revenue Experts. Published 19 November 2026. Reading time about 7 minutes. Doing time about 10 minutes for one page, 25 for five.
Is my main content in the raw page?
Short answer: in Chrome, press Ctrl+U to see the HTML your server sends before any script runs, and use Ctrl+F to search for your main heading, a price and one line from lower down. Then open DevTools, disable JavaScript and reload. If your text shows both times, a crawler that skips scripts can read it. If the page comes back blank, your content is built in the browser, where those crawlers never see it.
Every Thursday we send The Revenue Signal, our weekly newsletter: one study or how-to, with the numbers and the sources. Subscribe here: https://revenueexperts.beehiiv.com/?utm_source=academy&utm_medium=lab&utm_campaign=lab-4-raw-page
Why can nobody tell the board where AI gets its answer?
At the quarterly board meeting, a director reads out what ChatGPT says about the company when a buyer asks about our category. Some of it is right, some is thin, one line is wrong. The director asks a plain question: where did that answer come from, and what is it based on? Nobody in the room can say. The quieter follow-up is worse. Can ChatGPT, Claude and Perplexity even read our site, or are they answering from scraps? The board signed off a fast new website last month, and nobody checked what those tools actually see.
Here is why the question is hard to answer. In the head of marketing’s browser, every page is full of text: what the product does, who it is for, the pricing table. But a browser runs JavaScript, and several AI crawlers do not. The same page arriving at one of those crawlers can be a few lines of code and an empty box, something like a lone <div id=”root”></div>, with the product description, the pricing and the customer story all inside a script the crawler never runs. The page is reachable, indexable and fast, and it holds nothing to quote.
Nobody did anything wrong. The site was tested in a browser, and browsers run JavaScript. This Lab tests it the other way, and gives the board a yes or no for each page in about ten minutes.
What does the raw-page check look at?
The raw-page check, check 1.4 of the 30 checks we use to judge whether a page can be found and quoted by AI assistants, is in the first category, Retrieval Eligibility and Access. It is worth 4 of 100 points and it is scored by a fixed rule, not by a judgement call. Take the HTML exactly as the server sends it, remove every script, style, noscript and template element, and count the characters of text left in the body. 200 characters or more passes; fewer fails; a challenge page or an error counts as blocked. 200 characters is about two sentences. The check is not asking whether your whole page is built on the server, only whether a crawler that does not run scripts sees anything at all.
Do AI crawlers render JavaScript?
Vercel, which hosts a large share of websites built on modern JavaScript frameworks, measured AI crawler traffic on its own network and found that none of the major AI crawlers render JavaScript: OpenAI’s OAI-SearchBot, ChatGPT-User and GPTBot, Anthropic’s ClaudeBot, Meta’s crawler, ByteDance’s Bytespider and PerplexityBot [1]. In the month it measured, GPTBot made 569 million fetches and Claude’s crawler 370 million [1]. About 12 in 100 of ChatGPT’s fetches and about 24 in 100 of Claude’s were JavaScript files, downloaded and not run [1].
Vercel sells hosting for server-rendered sites, so read its recommendation with that in mind. The measurement is the useful part: if your text appears only after a script runs, these crawlers do not see it.
How do I view the page source? (Test 1, 3 minutes)
Open the page in Chrome. Press Ctrl+U on Windows or Linux, or Cmd+Option+U on a Mac, to display the page source [2]; or type view-source: in front of the address. A new tab shows the HTML exactly as the server sent it, before any JavaScript ran.
Press Ctrl+F (Cmd+F on a Mac) and search for the first five or six words of your main heading.
Search again for the first words of the paragraph under it.
Search for one price, one product name and one sentence from lower on the page, such as an FAQ answer or a customer story.
Write on the worksheet which of the four searches found the text. Found means a crawler that does not run scripts can read it. Not found means it arrives later, from a script.
How do I see the page with JavaScript off? (Test 2, 3 minutes)
This shows you the page the way those crawlers see it [3].
On the page, press F12 to open Chrome DevTools.
Press Ctrl+Shift+P (Cmd+Shift+P on a Mac) to open the Command Menu [3].
Type javascript, select Disable JavaScript, and press Enter [3].
Reload the page with DevTools still open.
To switch it back on, run Enable JavaScript from the Command Menu, or close DevTools.
Look at what is left. A full page of text is a pass. A blank page, a spinner, or only a header and footer means the content is built in the browser. Also open every tab, accordion and read-more panel: if a panel is empty with JavaScript off, its text is fetched on click and is not in the raw page.
How do I check more than one page? (Test 3, 15 minutes)
One page tells you about one template. Run Tests 1 and 2 on five pages and fill one row each on the worksheet:
| Page | Why this one |
|---|---|
| Home page | Often the most designed, sliders and animated sections |
| Your main service or product page | The page buyers’ questions should land on |
| Pricing or plans | Prices are often in a widget or a script-built table |
| One blog post or guide | Usually passes; a useful control |
| Any page built with a page builder, an app or a plugin | Where injected content hides most often |
Table 1. Five pages to test. A failure on one template usually repeats on every page built from it.
What does a good result look like?
Your main heading, the first paragraph, the price and at least one lower section are all found in view-source.
With JavaScript off, the page still reads as a page: headings, paragraphs and lists, even if the layout looks plain.
Tabs and accordions show their text with JavaScript off.
Five rows on the worksheet, dated, with the template each page uses.
The wrong reading to avoid: Google indexes our pages, so the crawlers must see the text. Google renders JavaScript for its index. The crawlers in Vercel’s study do not [1]. A page can rank in Google and still look empty to an AI search crawler.
What do I do if a page fails?
Most sites on a traditional platform pass this check by a wide margin. When one fails, the cause is usually one section, not the whole platform. This table is written to forward to whoever edits your site.
| Platform | What usually puts text outside the raw page |
|---|---|
| WordPress | Page-builder lazy or dynamic sections, sliders, AJAX-loaded Query Loops, tabs or accordions that fetch panels on click |
| Webflow | Script-built widgets; Code Embed HTML itself is in the page |
| Shopify | Product recommendations fetched on scroll, predictive search, app embeds that fetch their own data such as reviews and FAQ widgets |
| Wix | Asynchronous Velo code that does not finish before the server render; HTML embeds, which are sandboxed iframes |
| Squarespace | Summary blocks, galleries, embed blocks and code-block JavaScript |
| HubSpot | Embedded forms, the chat widget, and content a developer moved into JavaScript |
| React, Vue, Angular, Svelte in default mode | The whole page: the server sends an empty shell and a script builds everything |
Table 2. Where the text usually goes missing, by platform.
For a site built as a JavaScript application, there are three fixes, in order of preference. Server-side rendering: the server sends finished HTML for every request (Next.js, Nuxt, SvelteKit, Angular SSR, Remix, Astro); the most complete fix, and the most developer work. Static site generation: pages built as plain HTML at deploy time, good for marketing pages and blogs. Pre-rendering for crawlers: a service serves a snapshot to crawlers; the least work and the most fragile, so treat it as a bridge.
A first step any developer can do in an afternoon: put the main heading and the first paragraph into the HTML the server sends. That is enough to pass this check. It is not enough for the evidence checks later in category 2, which read far more of the page, so plan the full fix after.
What does this check not tell you?
Passing means a crawler that does not run scripts can read your page. It does not mean an assistant will quote it, name you or rank you. Those depend on what the text says, which is what Lab 3 and the later category 2 checks are about. And the four searches in Test 1 are a quick check, not a full count.
Where this fits
This is the raw-page check, check 1.4 of the 30 checks we use, in the category Retrieval Eligibility and Access. It is worth 4 of 100 points. The rest of the checks publish one at a time in the Lab: https://onlinemarketingacademy.ai/lab/
If a page comes back blank, first make sure the AI crawlers are even allowed to reach it. Lab 1 checks that in your robots.txt file: https://onlinemarketingacademy.ai/lab/ai-crawlers/
If supplier numbers reach your finance team without their source, that is the same visibility gap in the back office. The Desk shows the 3-invoice check: https://onlinemarketingacademy.ai/the-desk/check-invoices-against-contract/
When your pages pass and the next question is what ChatGPT, Claude and Perplexity actually say about you, that is the AI citation audit: 50 buyer questions, three assistants, three runs each, 450 saved answers, and a fix list with an owner next to every gap, in 2 to 5 days, for $1,495. We guarantee the deliverables, never the rankings. Book 20 minutes with John Bush: https://meetings.hubspot.com/john2956?utm_source=academy&utm_medium=lab&utm_campaign=lab-4-raw-page , or write to us through the contact page: https://revenueexperts.ai/contact/?utm_source=academy&utm_medium=lab&utm_campaign=lab-4-raw-page
Download the one-page worksheet (PDF): download the worksheet (PDF)
If this was useful, the next one arrives on Thursday. Subscribe to The Revenue Signal: https://revenueexperts.beehiiv.com/?utm_source=academy&utm_medium=lab&utm_campaign=lab-4-raw-page
Frequently asked questions
Does Google see my JavaScript content? Google renders JavaScript for its own index, so a client-rendered page can still rank in Google. The AI crawlers in Vercel’s study do not [1].
Is 200 characters really enough? For this check, yes: it is a floor, not a goal. The evidence checks later in category 2 read far more, so the more of your main content the server sends, the better.
Our site is on WordPress. Do we need to test at all? Usually WordPress passes. Test the pages built with a page builder, slider or tabs plugin; that is where text goes missing.
Will switching JavaScript off break anything? Only in your own browser tab while DevTools is open. Close DevTools and it is back to normal [3].
Who should own fixing this, and what will it cost? Your web developer, or whoever manages the site, owns the fix. The first step, putting the main heading and first paragraph into the HTML the server sends, is usually an afternoon. A full move to server-side rendering is a larger project, so scope it with whoever built the site. You do not need us to run the check or make that first fix.
We just launched a new site. Do we have to change it? Not always. Many new sites pass this check. Run the two tests first; if your text shows in view-source and with JavaScript off, there is nothing to change. The point is to check rather than assume.
Sources
[1] Vercel, “The rise of the AI crawler”, https://vercel.com/blog/the-rise-of-the-ai-crawler , 17 December 2024, opened 19 November 2026. Vercel sells hosting for server-rendered sites.
[2] Google Chrome Help, “Chrome keyboard shortcuts”, https://support.google.com/chrome/answer/157179?hl=en , opened 19 November 2026 (lists Ctrl+U on Windows and Linux and Cmd+Option+U on a Mac for “Display non-editable HTML source code for the current page”).
[3] Chrome DevTools, “Disable JavaScript”, https://developer.chrome.com/docs/devtools/javascript/disable , opened 19 November 2026.



