Scanora AI

I was building a website, and everything was fine. Then one day my performance score dropped from 100 to 70 and below. That matters more than it sounds: a performance score under 90 can hurt your traffic. Long loading times mean bad user experience. Bad user experience means fewer customers. Fewer customers is bad for you. I ran an audit on my own site with Scanora, the tool I'm building, and it flagged performance as the weakest area. My first instinct was to blame a <motion.div> - Framer Motion animations are an easy scapegoat. But digging deeper, the real numbers told a different story:

- TTFB: an excellent 19ms

- Total page weight: only 430KB

- No single large or slow resource

The actual gap was between server response and First Contentful Paint: 2.7 seconds for content to actually appear on screen. Fast server, slow paint - that's the signature of a render-blocking resource, not a slow backend.

The cause: my CSS was being loaded via a standard <link rel="stylesheet"> tag. That forces the browser to download HTML, parse it, discover the stylesheet link, then fetch it - before it can paint anything. One more network round-trip standing between a fast server and a visible page. It turns out Next.js shipped an experimental flag for exactly this in 16.3.3: inlineCss.

// next.config.js

const nextConfig = {

experimental: {

inlineCss: true,

},

}

With it enabled, Next.js swaps the <link> tag for an inline <style> tag — your CSS arrives with the HTML itself, in the first response, no extra request required.

Have you ever wondered if you're getting leads through ChatGPT, Claude, Perplexity, or Gemini? I think tracking this is more important than ever more and more people are asking their AI assistant instead of googling.

And now that you can even advertise through ChatGPT, there's an obvious next question: how do you actually know if that ad is working?

That's exactly why I built AI-referral lead tracking into Scanora a small snippet that tells you which AI model actually sent you a lead, instead of guessing.

Has anyone else noticed AI-driven traffic showing up in their own leads yet and if so, how are you currently tracking it?