CrawlCheck

Findings · 2026-09-30 · By

Our extension read a page as 36.5% JavaScript. It was zero.

The CrawlCheck browser extension overstated how much of a page needed JavaScript, for two reasons that both pushed the number up. Version 1.0.2 fixed both.

On 25 September 2026 the CrawlCheck browser extension reported 36.5 percent of a homepage as dependent on JavaScript. The true figure was zero: the same 2,850 words appeared in the raw HTML and the rendered page. Two causes, both inflating: punctuation written as HTML entities in the raw copy did not match the decoded characters in the browser, and the WordPress admin bar shown to a signed-in owner was counted as page text. Version 1.0.2 fixes both, and test pages read 49.9 and 100 percent as expected.

Reported JavaScript-only share at each step
as shipped in 1.0.136.5%
after full entity decode11.9%
after admin bar removed0%
test: injected paragraph49.9%
test: empty shell, all text by script100%

The extension compares the raw HTML a server sends with the page a browser shows after JavaScript runs. The text that only appears after rendering is the share a crawler that does not run JavaScript never sees. On 25 September it reported that share as 36.5 percent for a homepage on a site we run. The true figure was zero: 2,850 words in the raw HTML, the same 2,850 words in the rendered page.

What went wrong? #

Punctuation spelled two ways. Raw WordPress HTML writes curly apostrophes, dashes and arrows as entities such as ’. The browser shows the characters themselves. The extension did not decode all of these, so every sentence with an apostrophe looked different in the two copies and was counted as appearing only after rendering. Decoding every entity took the reading from 36.5 percent to 11.9.

The person testing was signed in. WordPress adds an admin bar for a signed-in owner. It is added by script and never served to visitors or crawlers, but the extension counted it as page text. Removing it took the reading from 11.9 percent to zero.

StepReported JavaScript-only shareCause
as shipped in 1.0.136.5%entities not decoded in the raw copy
after full entity decode11.9%signed-in admin bar counted as page text
after admin bar removed0%2,850 of 2,850 words identical
test: injected paragraph49.9%correct
test: empty shell, all text by script100%correct

Why was the wrong number believable? #

Both errors could only push the reading up, and a third of a page depending on JavaScript is a normal result on many modern sites. So the number looked like a finding, not a fault. It was caught only because we already knew the page was fully server-rendered. The lesson we keep relearning, as with the image check that counted placeholders, is to test a new reading on a page whose true value you know before trusting it on one you do not.

What does it mean for your own reading? #

If you use any tool that compares raw and rendered text on a WordPress site, sign out first or use a private window, so the admin bar is not counted. From version 1.0.2 the CrawlCheck extension handles both issues itself. It is still one browser at one moment, so a page that renders differently by screen size or session can read differently. For the crawler side, see do AI crawlers render JavaScript.

Live, as you read this: the corpus now holds 570,959 domains across 9,163 scans. The figures in this piece were measured on the date above; this line is not.

CrawlCheck’s dataset when this page loaded: 570,959 sites measured.

Share this finding:LinkedInXEmailFollow Emmanuel on LinkedIn

Questions this post answers

What does the extension's JavaScript-only share measure?

The share of a page's visible text that appears in the rendered page but not in the raw HTML the server sent, which is the share a crawler that does not run JavaScript never sees.

Why did it report 36.5 percent on a page that needed no JavaScript?

Punctuation written as HTML entities in the raw copy did not match the decoded characters in the browser, and the WordPress admin bar shown to a signed-in owner was counted as page text.

What changed in version 1.0.2?

It decodes every entity in both copies, ignores differences that are only punctuation, and removes the admin bar before comparing.

How was the fix tested?

On the original page, which now reads zero, and on two test pages: one with a paragraph added by script, which reads 49.9 percent, and one whose whole body is written by script, which reads 100 percent.

Should I sign out before testing my own WordPress site?

Yes, with any raw-versus-rendered tool. The admin bar is only shown to signed-in users and is not what visitors or crawlers receive.

Related findings

The main product

Found this on your own site? We fix it for $100.

Scan free to see where you stand. The fix is one site, every finding implemented and re-measured, with a sealed before and after.

Free browser extension

Check this on any site you open, without leaving the page.

CrawlCheck for Chrome grades every site you visit A–F and shows, from your own browser, what search engines and AI assistants are actually served. Free, no account.

Write for the library

Measured something yourself? Submit an article.

Send your finding, the data and the method. A person reads every submission and replies by email; accepted articles are published under your name with a link to your study. Free to submit. Articles are accepted on merit; no one can pay to have one published.

Submit an article

How it works · a reply within about a week

Spotted a mistake or a bug on this page? Report an error. Confirmed errors are fixed and logged on the corrections page.

Comments

Comments are read before they appear. Nothing is published automatically, and no account is needed. An AI agent with a published signing key can post without the queue: it signs the request and the signature is checked against its own domain (how). Those appear at once, marked verified.

Writing about this? Facts, live figures and marks — every number on that page is dated and traceable to a scan.

All findings · The dataset · How the dataset works