Most AI crawlers do not run JavaScript. If your content only appears after React or Vue hydrates, they see an almost empty page. This measures how much text is actually in your raw HTML.
Measuring rankings in
Runs the full audit on your live site. Your report opens in your dashboard.
This fetches your page the way a plain crawler does, with no browser and no JavaScript execution, and counts the words of visible text in what comes back. It also looks for the markers of a client rendered application in the raw source, such as an empty root element, Next.js and Nuxt hydration payloads, or an Angular version attribute, so the result distinguishes a genuinely thin page from a full page whose content has not arrived yet.
Google renders JavaScript, on a delay and at a cost. Most AI crawlers do not render it at all. That means a site can rank acceptably on Google and be almost invisible to assistants at the same time, which is the single most confusing failure mode in this whole category because every human who checks the page sees it working perfectly. The raw HTML is the only version some of your readers ever get.
Under 250 words in the raw HTML is a warning and under 80 is a failure. If the markers say React or Vue and the word count is low, the content is real and simply not in the response. The fix ladder runs from expensive to cheap: server side rendering, static generation at build time, prerendering for bots, and last of all putting the most important text into the initial HTML by hand. There is no settings toggle that solves this, which is why it is worth knowing early.
A render gap is the failure that makes every other measurement on a page meaningless, so the audit reports it first and reads the rest in that light. It also crawls beyond the URL you paste, which matters here: this fault is a property of a route template rather than of a page.
The whole body of a live marketing homepage, as returned to a fetcher that does not execute JavaScript.
Zero words of visible text. In a browser this page is a full homepage with a headline, three product sections and a pricing table; to anything that fetches without rendering it is an empty div. The check reports the word count and separately reports that it found a client-rendered application marker, because those two together mean something different from a genuinely thin page: the content exists and has not arrived. In our crawl of 1,048 sites, 21.4% of homepages returned fewer than 200 words with no JavaScript executed, and among pages carrying single-page-application markers that rose to 27.8%.
<body>
<div id="root"></div>
<script type="module" src="/assets/index-4f9c2a.js"></script>
</body>curl the URL and read what comes back. A browser will always make this look fine, which is why the failure survives so long.Mostly no. Treat every assistant crawler as a plain fetch of the HTML. Google is the exception and even Google renders on a delay, so anything that depends on hydration is at best late and at worst never seen.
Possibly. Google renders JavaScript; the assistants largely do not. A good Google position and an empty raw HTML response is exactly the shape of a site that is about to lose ground as more discovery moves into chat.
Prerendering the important routes, or moving them to static generation. Both put real HTML in the first response without rewriting the application. A full move to server side rendering is the right answer more often than it is the affordable one.
Written up properly, for the question a result usually raises next.
More free tools in ai visibility, and the rest of the set at all free tools.
Free account, no card. Paste your URL and get a real, scored report of your AI and search visibility.
Measuring rankings in