What "audit your own site" actually finds

We've run 800+ audits on the public /try endpoint. Here is the distribution of what we find — what most sites are missing, what almost no sites have, and what the highest-leverage fix is.

Merkava's free audit at withmerkava.com/try has been run on 800+ public domains. The audit checks 30+ SEO/GEO surfaces and returns a gap list. Here's what we find at scale.

What almost no sites have

Three signals show up missing on more than 80% of sites:

• /llms.txt — missing on 91% of audited sites. The file most operators have not heard of.
• ai-content-policy meta tag — missing on 87%. Sites either silently allow AI crawlers or block them entirely; the explicit policy is rare.
• WebSite + Organization JSON-LD on the homepage — missing on 78%. Especially common on sites built before 2022 or on no-code platforms that don't expose the head tag.

These are the highest-leverage fixes because the floor is so low. Adding them puts you ahead of 80%+ of competing sites for AI-search citations.

What most sites have but get wrong

Three signals are present but commonly malformed:

• FAQPage schema on /pricing. Present on 41% of sites with a /pricing page, but only 12% of those have buyer-intent questions. Most are filled with marketing-team questions ("Why choose us?") instead of buyer questions ("How much does this cost?").

• Open Graph completeness. Present on 67% of sites, but only 39% have all four (og:title, og:description, og:image, og:url). The most common gap is og:image — a meta tag references an image URL that returns 404 or is the wrong dimensions.

• Canonical link tags. Present on 71% of sites, but only 54% have them on every key page. The pattern is: homepage and product pages are tagged, blog and resource pages are not.

What most sites have but is stale

The two most common stale signals:

• sitemap.xml. Present on 88% of sites, but the median age of the last update is 4 months. New pages aren't getting crawled because the sitemap isn't being regenerated.

• /llms.txt content (when present at all). Of the 9% of sites that have one, ~40% are stale — they list pages that no longer exist or are missing the last 6+ months of content.

The 1% that have everything

About 1% of audited sites pass all 30+ checks. The pattern: they're either run by a senior SEO professional with a developer relationship, or they're recent builds where the developer was thoughtful from day one.

For the other 99% — there's room to ship.

The highest-leverage fix

If you can only do one thing, do /llms.txt + ai-content-policy meta + WebSite/Organization JSON-LD. That's three signals, ~30 minutes of work, and it puts you ahead of 80% of competing sites.

After that, FAQPage on /pricing with real buyer questions. ~45 minutes if you have the answers ready.

After that, make sure your sitemap regenerates automatically on every deploy.

Everything else is incremental.

What an audit actually looks like

The audit at withmerkava.com/try returns a ranked gap list, the exact fix content for each gap, and a sharable URL so you can send it to whoever owns the fix. ~5 seconds, no signup, no credit card.

If running the fixes yourself feels like work you don't want, Merkava's SEO agent (Beacon) ships them as PRs against your repo on a weekly recheck cadence.