We Ran Our New Deep Audit on Moz, Ahrefs, Semrush — and Ourselves

Before we let anyone else use Deep Audit, we did the only honest thing: we pointed it at our own homepage — and at the three biggest names in SEO software. No cherry-picking. Here's exactly what came back, what we fixed, and what it taught us about our own product.

First, the leaderboard

We ran a Deep Audit on the homepage of each site (Ultra tier, single render). The headline LAI score is out of 5; everyone landed in the same "good, not great" band:

Two things jumped out immediately.

Nobody aces it. The entire industry — companies whose whole business is ranking — clusters within a third of a point, and not one breaks into "excellent."

Everyone shares the same blind spot: structured data. Section 6 (Schema) was the lowest-scoring section for all four sites (2.2–2.6 out of 5). Every single one was missing schema types it could reasonably use — Service, Review, AggregateRating, and more. The companies that literally publish the guides on structured data don't fully implement it on their own homepages. In an era where AI answer engines lean hard on structured data to understand a page, that's a real, shared gap.

There's also a telling paradox in the sub-scores: classic on-page SEO was every site's weakest metric (2.3–2.7), while AI-readability was strong (3.6–4.0). These sites are built for humans and, increasingly, for LLMs — but they leave classic structured SEO signals on the table.

Then we ate our own cooking

We came third. So we did what we'd want any customer to do: we opened the findings and fixed what was real. Deep Audit flagged two genuine problems on our homepage:

1. Missing structured data. We were missing schema types we legitimately should have — Service, SiteNavigationElement, and an Organization contact point. (We deliberately did not add Review or AggregateRating schema — we don't have real reviews to cite, and faking them is schema spam that Google penalizes. That restraint matters, and Deep Audit doesn't push you to fabricate.)
2. No Content-Security-Policy. A basic security header we'd simply never set.

We added the JSON-LD and a proper CSP (plus a few companion security headers), redeployed, and re-audited. The result:

An afternoon of work — no re-platform, no content rewrite — moved our overall score from mid-pack to the top of that leaderboard. Just acting on what the audit surfaced.

The most valuable thing it found was a bug in our own tool

Here's the part we're proudest of. While reading our own findings, we noticed the AI kept hedging on structural questions — "no <main> element visible in the provided excerpt" — about a page that clearly has a <main> element. It was marking real, present things as "can't verify."

Digging in, we found the cause: the analyzer was handing the AI a tag-stripped text excerpt of the page, then asking it structural questions the stripped text couldn't answer. Classic garbage-in. We fixed it — the AI now receives an authoritative, DOM-derived structural summary — and those false negatives disappeared. Our Content & Entity score jumped a full point (3.08 → 4.08), not because the page changed, but because the tool finally saw it correctly.

Every audit you run is sharper because we audited ourselves first.

The honest caveats

We'd rather tell you the limits than have you find them:

• These are single-run, homepage-only audits. A different page or a re-run can move AI-scored sections a little.
• Schema scoring is deliberately strict, but a mature, well-built page (like most of these) is already near the ceiling on the technical sections — the wins came from the two genuine gaps, not from gaming a number.
• The competitor scores are a point-in-time snapshot. This isn't a takedown of Moz, Ahrefs, or Semrush — they're excellent products, and the fact that everyone clusters together is the actual story.

Why this is worth paying for

Here's what this exercise actually demonstrates — and why Deep Audit earns its place next to (or ahead of) the tool you're paying $100–$200/mo for today:

• It graded us for the AI-search era. The biggest lever it found wasn't a keyword or a backlink — it was structured data and how machines read the page. That's the axis the classic suites don't score, and it's where discovery is going.
• It gave us fixes, not a wall of red. Two clear, real problems with plain-English evidence — the kind of prioritized, honest output you'd expect from a consultant, not a crawler.
• It runs in our pipeline, not a dashboard. We re-audited automatically after each deploy. Try wiring Moz or Semrush into your CI/CD like that.
• It didn't tell us to cheat. It flagged missing Review schema — and we chose not to add it, because we have no real reviews to cite. A tool that rewards fabricating signals isn't measuring your SEO; it's gaming its own number.

All of that for $39–$99/mo, as an API you can automate — not another seat in a dashboard.

Try it on yourself

That's the whole pitch: run Deep Audit on your own worst-performing page, fix the two or three things that are genuinely broken, and watch the number move. It's a Pro/Ultra feature and your existing key already works.

import seoscoreapi as seo
before = seo.deep_audit("https://yoursite.com", API_KEY)
print(before["scores"]["lai_score"], "— now go fix the top findings")

Read the announcement → · How-to guide →