NEW:Turn AI visibility insights into ChatGPT ad creative.

Vercel setup

Decide whether Vercel already delivers crawler-readable HTML, then add narrow middleware only when it is needed.

On this page

Check the delivered HTML before adding middleware

Vercel can host static sites, server-rendered applications, and hybrid frameworks. Do not assume that every Vercel deployment needs prerendering. First inspect the initial response for a representative public page.

If the response already includes the title, headings, links, and meaningful text, focus on normal technical SEO checks instead. If it contains only a client-side shell, a crawler-specific rendering path may be appropriate.

Add narrow middleware only when necessary

Middleware should run only for public document requests and a defined set of crawler user agents. Exclude static assets, data routes, API routes, authentication, dashboard, checkout, and user-specific pages.

Keep the render-service credential in Vercel environment variables and call it server-side. Do not expose it through a public environment variable or client-side fetch.

Prevent loops and mismatched hosts

Use a separate origin or a route that does not re-enter the middleware path. Ensure the production domain, deployment URL, canonical URL, and redirects are intentional. Test the apex and www hostname separately if both are active.

Validate the result

Compare standard and crawler-style HTTP responses for the same URL. A crawler response should be complete, public, and consistent with the visitor-visible page. Record the status and headers before considering the configuration complete.

Instructions for your connected site

Open account setup for domain-specific values, diagnostics and copy-ready configuration. These stay private to your workspace.