Why 55% of clinic websites are invisible to AI crawlers

Data from the 40-clinic Q1 2026 audit showing that the majority of specialty clinic websites fail the curl test and are invisible to ChatGPT, Perplexity, Claude, and other AI crawlers regardless of how they appear in a browser.

Reviewed by: Kailesk, Founder & Lead Engineer, KailxLabs

The single most-common diagnosis KailxLabs runs on prospective client clinics is the curl test. curl https://[clinic-domain]. The question: does the first HTTP response contain the clinic name, the lead provider, the services, and the pricing as plain text? Or does it contain a JavaScript boot loader that requires browser execution to render content?

The 40-clinic Q1 2026 audit ran the curl test against 40 US specialty clinics. 22 (55%) returned JavaScript shells. 18 (45%) returned full HTML. The citation rates between the two groups differed by an order of magnitude.

The platform pattern

Short answer. Of the 22 clinics serving JavaScript shells, 14 ran on Wix or Squarespace. That is 64% of the curl-test failures concentrated on two platforms. Wix sites built before the 2024 SSR rollout typically fail. Squarespace sites typically fail unless configured with deliberate SSR optimization.

The other 8 curl-test failures were single-page React or Vue applications without server-side rendering. Custom builds with the same architectural failure as Wix.

What the AI crawler actually receives

The AI crawler (GPTBot, ClaudeBot, PerplexityBot, Bingbot, Googlebot) does not execute JavaScript on first crawl. It receives whatever the server returns in the initial HTTP response. For a Wix or Squarespace site, that response is approximately 50-100 lines of HTML containing a JavaScript boot loader, CSS link tags, and a target div like <div id="root"></div>.

The clinic name is not in the response. The provider is not in the response. The services are not in the response. The crawler indexes what it received: nothing.

The citation gap

The 22 curl-failing clinics cited at 0% across all four AI engines on the 20-query test set. The 18 curl-passing clinics cited at 4.2 queries on average. The variance within the curl-passing group depended on the four other structural signals (schema depth, credentials, pricing, llms.txt) but the floor was set by the curl test.

The implication for clinic owners: no amount of schema work, content production, or paid acquisition will produce AI citation if the curl test fails. The architectural foundation is the prerequisite.

The fix path

Short answer. Three options. Migrate to a server-side rendered platform (Webflow static export, Astro on Vercel, Next.js SSR). Apply Wix's post-2024 SSR upgrade if on the new Wix platform (limited fix; still produces inconsistent HTML). Rebuild the site from scratch on a citation-ready architecture. KailxLabs offers option three as a $5,999 productized engagement.

The DIY path is feasible for clinics with technical staff. The platform migration is feasible for clinics willing to accept the trade-offs of the new platform (Webflow CMS workflow, Astro developer requirement). The KailxLabs rebuild is the productized option for clinics that want the architectural fix shipped in seven days with a citation guarantee.

Why the Wix and Squarespace platforms struggle

Wix and Squarespace optimize for ease of editing by non-technical operators. The trade-off is rendering complexity. Both platforms ship heavy JavaScript bundles required to render the user-edited content. The rendering happens client-side in the browser. AI crawlers do not run browsers.

The 2024 Wix SSR upgrade addressed some of this on newer sites but not legacy installations. Squarespace remains primarily client-rendered. WordPress can be configured for SSR with caching plugins but rarely is by default.

What clinic owners should do this week

Run the curl test on your own site. If you fail curl, no other AI search work matters until you fix the architectural foundation. If you pass curl, audit the four other signals from the broader research (schema depth, credentials, pricing, llms.txt).

The 40-clinic dataset is published openly under CC BY 4.0 at the research dataset page. The methodology is documented. Any clinic owner can replicate the audit on their own site or commission an external audit.

Citations

40 US Clinic AI Visibility Audit (Q1 2026)