The render gap
Modern websites assemble their text in the browser. For people that works. For Claude it does not: in a measurement with eight test scenarios, Claude managed none of them. The page is then not badly ranked — it is empty.
The answer in one sentence
If your text only comes into being through JavaScript, Claude fetches the page, gets an empty shell and cites nothing. No content in the world helps against a fetch that finds nothing.
The measurement
searchVIU built eight typical cases in which content only appears through JavaScript and checked how many of them the systems manage on a live fetch.
An important distinction
These figures apply to the live fetch — that is, when a system fetches the page at the moment somebody asks. Google AI Overviews and Microsoft Copilot work differently: they fall back on the normal search index, and Googlebot and Bingbot render JavaScript. For those two the question is moot.
That is why we only warn where it applies. A warning for Google based on a measurement for Claude would be an alarm without a cause — and it is exactly such warnings that make people stop looking.
How you can tell
The simplest test takes ten seconds: open your page, switch JavaScript off in the browser and reload. Whatever you can still read is what Claude sees. Whatever disappears, Claude does not see.
Our quick check does the same automatically: it compares the source with the rendered version and tells you per surface how many of your pages really arrive — as a figure, not as an assessment.
What follows from this
Important content belongs on the server side — as finished HTML, not as assembly instructions. For most site builders and content management systems that is the norm; it becomes critical with single-page applications and with content that is only loaded after scrolling.
Stand: 1. September 2026 · Messung: searchviu.com/en/ai-crawlers-javascript-rendering