Scanora AI

I've been thinking a lot about how to land my first customers, and here's an idea I ended up building. Our tool does free website audits - no signup, no email, nothing. You just enter your URL, hit "Check now," and it crawls your site and gives you SEO, Performance, and GEO scores (GEO = how visible you are to ChatGPT, Claude, Perplexity, Gemini and Google AI Overview) in under a minute. As an anonymous visitor you only see a preview of the issues we found. Creating a free account unlocks the full list - every issue across SEO, Performance, GEO, plus the keyword analysis. Still, at that point you've only seen scores and problems, not what our "full report" experience actually looks like end to end (the prioritized fixes and AI-generated action plan are part of our paid plan).

That's a bit of a leap of faith for someone who's never seen the real thing. So I published one: I ran our own website, scanora.ai, through the tool on the paid tier and made that complete, unedited report available as a free download - same depth, same format, everything a paying customer gets. Just from our own site instead of theirs.

The idea is simple: instead of asking people to trust a description, let them see the actual product before they commit to anything.

Still early days - curious what you all think: is "show, don't tell" enough to get someone from curious visitor to first audit, or does it need more?

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?