The technical SEO audit tool that checks your fixes worked

It checks the parts of your site search engines depend on before content quality matters at all: status codes, redirects, links, indexability, metadata, speed, sitemaps and structured data. LogNorm's audit covers these with 29 rules in six categories, each with a severity.

Every rule says why it matters and how to fix it. A slow page, for example, is one whose HTML took more than 2.5 seconds; a deep page is more than 3 clicks from the home page.

In two tiers. First an inventory lists every URL LogNorm can find in your sitemaps, feeds and links; then a deep read fetches and reads the most important pages first, such as the home page, navigation, commercial pages, recently changed and shallow pages.

Each run continues where the last stopped, reading pages never read or changed since before stale ones, so coverage grows run by run. Pages per run depend on your plan: 100 on Free, 1,000 on Starter, 3,000 on Growth and 10,000 on Scale. When more pages are waiting, LogNorm asks whether to read within your plan, read the rest with extra credits, or upgrade. Pages that haven't been read are never audited, and every audit states how much of the site it covered.

A finding reports only what was actually checked. Where LogNorm couldn't verify something, the check is listed as Not checked instead of passing, and where AI made the call, the finding says so.

• Noindex is confirmed per page. LogNorm re-fetches the page and keeps the robots meta tag and the X-Robots-Tag header apart. For header-only cases it also checks whether Google has the page indexed.
• Link-graph rules need coverage. Orphan and deep page rules only run when at least 90% of known pages are analysed. Below that they show as Not checked, never as passing.
• AI-judged rules are labelled. Thin content and possible keyword cannibalisation need judgement, so their findings carry LogNorm's call with a confidence percentage.
• Every finding keeps its URL, so every issue can be exported and checked by hand.

Findings are grouped by rule, not listed page by page, so a site-wide problem reads as one card. Filter by Critical, Warning or Notice; each card shows why it matters, how many pages are affected and how many are already fixed.

For broken internal links and links pointing at redirects, the card names the URLs behind most findings, such as a broken footer link on every page, so you fix one link instead of chasing thousands of findings. The audit score starts at 100 and each failing rule costs points by severity, scaled by the share of pages it affects. Export all issues (CSV) downloads every open finding with its URL, severity, detail, how to fix it and whether it was measured or AI-judged.

Each failing rule becomes one move, not one per page. Critical and warning rules become fix moves automatically after every audit, grouped so one ticket fixes every affected page; notices become moves when you click Create move.

Those fixes are ranked against the rest of your backlog, articles and commercial pages included, so a broken template that costs you hundreds of pages comes ahead of a missing alt attribute. See Moves and ranking for how the order is set and how to steer it.

Yes. Connect Claude Code, Codex or Cursor in the codebase that builds your site, and it can take audit moves from start to finish.

1. 1
   Claim the move

   The agent claims a fix from this week's plan, so two agents never work the same one.

2. 2
   Read the finding

   It reads the rule and every affected URL, then finds where those pages come from in your repo.

3. 3
   Fix the cause

   It changes the template, metadata, redirects, robots.txt or sitemap, not each page by hand, and comments what it changed with file paths.

4. 4
   Ship it

   Through your own commits or a pull request your team reviews, merges and deploys.

5. 5
   Ask for a re-check

   LogNorm re-checks the live pages. The move closes when everything passes.

Agents work on your own Claude or Codex plan, so their work costs no LogNorm credits; only the re-check LogNorm runs does. More on the AI SEO agent page.

Validate fix re-checks one rule on your live site without a full re-crawl. LogNorm re-fetches the affected pages, plus link targets for link rules and the other pages for duplicate rules, and re-runs that rule; robots and sitemap rules re-check the file.

Up to 1,500 affected pages are re-fetched per validation. The card shows how many pages are fixed, still failing or not checked. Fixed findings drop out of the audit, the CSV and the move; when everything is fixed the move is done, otherwise it updates to what's left. Orphan and depth rules depend on the whole link graph, so they need a re-analysis and a new audit instead. A fixed move then enters measurement against its Search Console baseline at 28 and 90 days.

Yes. Per website, the crawl and SEO audit can re-run daily, weekly or monthly on a paid plan or during the trial, and each run uses the same credits as a manual one.

The next run counts from the last one, so a manual run pushes it out. If a run can't start, for example because credits ran out, the schedule shows why. Score over time charts every audit, so you can see fixes land. For what AI crawlers and agents can read, the separate GEO audit runs alongside it.