How we made LumaRank’s own site pass our audit

We build a tool that tells brands how to be visible to AI, so it would be a bad look if our own site quietly failed our own audit. At first, it did. Here is exactly what we changed to fix that, written as a checklist you can run on your own domain.

The starting point: invisible by design

LumaRank began as a pure dashboard app. Almost everything lived behind authentication, which is fine for a product but means crawlers and AI models saw essentially nothing. So the first move was the biggest one: build a real public surface (this marketing site, docs, a help center and this blog) so there is genuine, crawlable content for search engines and AI to read in the first place. You can’t be cited for content that sits behind a login.

We shipped every AI signal we recommend

Then we put our money where our audit is and implemented the full checklist we score customers on:

• A real /llms.txt and /ai.txt, kept in sync with the rest of the site so they never drift.
• An AI-friendly robots.txt and a complete sitemap.
• JSON-LD structured data so models can attach the right facts to our brand.
• OpenGraph and Twitter cards, an RSS feed, and a web manifest.

And the SEO fundamentals underneath

None of the AI signals matter if the basics are broken, so we also covered full metadata with canonical URLs, one clear H1 per page, a sensible heading hierarchy, descriptive alt text, real internal linking, and static prerendered pages for fast Core Web Vitals.

The honest part

Dogfooding only counts if you are honest about the gaps. A couple of signals, like a public API, are still on the roadmap, and we won’t fake them: where something isn’t shipped yet, we say so rather than dressing it up. Everything else is live and verifiable.

What auditing ourselves caught

The real payoff came from an unexpected direction. Running our own audit surfaced a handful of false positives and edge cases in how we score a site, the kind of thing you only notice when you turn the tool on something you know inside out. We fixed them, which made the product more accurate for every customer, not just for us. That is the whole point of eating your own dog food: the problems you hit on yourself are usually the ones your users were hitting silently.

If you want to run the same playbook, start with getting started and audit your own domain. The list above makes a decent scorecard to work down.