Does it matter if my content only appears after JavaScript runs?
Yes, and it matters more for AI retrieval than for classic search. Many AI fetchers request the HTML and read whatever comes back without executing scripts. Content injected by JavaScript — accordion answers loaded on click, review widgets, service-area lists pulled from an API, client-side routed pages — can be entirely invisible to them while looking perfect in a browser. If it is not in the served HTML, assume the assistant never saw it.
The ten-second test
Open the page, view source, and search for a sentence you actually care about being quoted. Not a word from the navigation — a sentence from the answer. If it is not in the source, it is not in the document a non-rendering fetcher receives.
Do the same from outside the browser with a plain request, since some content only appears after a consent banner or a personalization script has run. What the fetcher gets is what matters, and the fetcher is not clicking anything.
The patterns that bite service business sites
- Accordion FAQs that load on click. The questions are in the HTML; the answers are not. This is the single most common version of the problem, and it hits exactly the content most likely to be cited.
- Third-party review widgets. Reviews inside an iframe belong to another document. They are not your page's content, no matter how they look.
- Service areas fetched from an API. The cities you serve are the most retrieval-relevant facts on a local page, and they are frequently the ones rendered client-side.
- Client-side routed location pages. Every branch shares one HTML shell, so every branch reads as the same page.
- Consent walls that replace the body. A blocking overlay can leave nothing behind it to read.
Why classic search was more forgiving
Conventional search crawlers have spent years building rendering pipelines, so a script-dependent site often ranks acceptably. That success is misleading. AI fetchers optimize for speed and volume, and a meaningful share of them do not render at all or give up quickly if the content is not there.
So a site can rank fine and be functionally invisible to assistants at the same time. That gap is one of the real differences between classic SEO and optimizing for AI answers, and it is a build decision, not a copywriting one.
The fix is boring and permanent
Server-render the substance and use JavaScript to enhance it, not to deliver it. Accordions should hide visible content with CSS rather than fetch it on demand. Lists of facts should be in the markup. Anything that must live in an iframe should also exist as text on the page.
Rebuilt this way once, the page stays readable to every fetcher that ever shows up, including ones that do not exist yet. That durability is why we treat it as part of how a site is built rather than a periodic optimization pass.
Topics: javascript · rendering · technical · diagnostics
Have a version of this question about your own business?
The useful answer usually depends on which systems you run and how they're connected. That's a conversation, not a blog post.