SEO audit checklist: the 48 checks of the Organiq site audit, with severity and fix

Updated November 10, 2026 · Organiq · third-party facts carry source and date in the text

Organiq (organiqgeo.com) runs 48 checks in its site audit: 18 errors, 22 warnings and 8 notices. This page lists each one with its code, what it checks and how to fix it, in tables generated from the code that runs the audit.

Which checks are errors, and bring the health score down?

17 checks are page errors. A page with any of them counts as a page with an error in both health scores. Each row is one check, with the code the tools use.

Code	What it checks	How to fix
`http_4xx`	Page answers with a 4xx error	Restore the page or redirect the URL to the equivalent page.
`http_5xx`	Page answers with a 5xx error	Fix the server error behind this URL.
`fetch_failed`	Page did not load	Check DNS, certificate and response time for the URL.
`broken_internal_link`	Internal link to a page with an error	Point the link to a URL that answers 200 or remove the link.
`redirect_loop`	Redirect loop	Make the URL end at a final destination that answers 200.
`title_missing`	Title missing	Write a unique <title> for the page.
`title_duplicate`	Title repeated on other pages	Give each page its own title.
`canonical_broken`	Canonical points to a page with an error	Point the canonical to a URL that answers 200.
`canonical_to_redirect`	Canonical points to a URL that redirects	Point the canonical to the final URL, with no redirect.
`noindex_in_sitemap`	Noindex page is in the sitemap	Remove the page from the sitemap or remove the noindex.
`sitemap_url_not_200`	Sitemap URL does not answer 200	Replace the sitemap URL with the final URL that answers 200.
`robots_blocked_in_sitemap`	Sitemap URL blocked in robots.txt	Allow the URL in robots.txt or remove it from the sitemap.
`noindex_with_clicks`	Noindex page receives clicks	Confirm the noindex is intended; it removes the page from results.
`mixed_content`	http resource on an https page	Load images, scripts and styles over https.
`hreflang_invalid`	Hreflang with invalid code or URL	Use valid language codes and absolute URLs in hreflang.
`hreflang_no_return`	Hreflang without a return link	Make the alternate page point back to this page.
`jsonld_invalid`	JSON-LD with invalid syntax	Fix the JSON in the ld+json block.

Which checks are warnings?

20 checks are page warnings. They are listed with the pages they affect and stay out of the health scores.

Code	What it checks	How to fix
`title_too_long`	Title over 60 characters	Shorten the title to 60 characters or fewer.
`title_too_short`	Title under 10 characters	Describe the page in the title with more words.
`description_missing`	Meta description missing	Write a meta description for the page.
`description_duplicate`	Meta description repeated on other pages	Write a separate description for each page.
`description_too_long`	Meta description over 160 characters	Shorten the description to 160 characters or fewer.
`h1_missing`	H1 missing	Add an H1 that states the page topic.
`h1_multiple`	More than one H1	Keep a single H1 and use H2 for the other headings.
`low_word_count`	Under 200 words of text	Check that the page answers the search with the text it has.
`img_alt_missing`	Image without alt attribute	Add alt to each image; use empty alt on decorative ones.
`redirect_temporary`	Temporary redirect (302, 303 or 307)	Use 301 or 308 when the move is permanent.
`redirect_chain`	Chain with more than one redirect	Redirect the URL straight to the final destination.
`internal_link_to_redirect`	Internal link to a URL that redirects	Update the link to the final URL.
`orphan_page`	Page with no internal links pointing to it	Link to the page from related pages.
`canonical_missing`	Canonical missing	Add <link rel="canonical"> pointing to the URL itself.
`not_in_sitemap_with_impressions`	Page with impressions missing from the sitemap	Add the page to the sitemap.
`slow_response`	Server response over 3 s	Reduce the server response time or use caching.
`html_too_large`	HTML over the size read	Reduce the page HTML; the rest was not analyzed.
`lang_missing`	lang attribute missing on <html>	Set the language, for example <html lang="en">.
`viewport_missing`	Meta viewport missing	Add <meta name="viewport" content="width=device-width, initial-scale=1">.
`nofollow_internal`	Internal link with nofollow	Remove nofollow from internal links.

Which checks are notices?

6 checks are page notices, the lowest severity. They are listed and stay out of the health scores.

Code	What it checks	How to fix
`deep_page`	Page more than 4 clicks from the home page	Bring the page closer to the home page with internal links.
`url_too_long`	URL over 115 characters	Shorten the page address.
`og_missing`	Open Graph missing	Add og:title, og:description and og:image.
`anchor_empty`	Internal link with no anchor text	Write text that describes the link target.
`jsonld_missing`	No JSON-LD structured data	Add the JSON-LD type that describes the page.
`canonical_multiple`	More than one canonical	Keep a single canonical tag.

Which checks look at the whole site?

5 checks look at the site and not at one page: robots.txt, the sitemap, llms.txt and AI crawlers. Their severities are mixed, including 1 error, and none enters the health scores; the table shows the severity of each.

Code	Severity	What it checks	How to fix
`ai_bots_blocked`	warning	AI crawlers blocked in robots.txt	Allow the AI crawlers you want to read the site.
`llms_txt_missing`	notice	llms.txt file missing	Publish /llms.txt if you want to point AI assistants to pages.
`sitemap_missing`	warning	Sitemap not found	Publish a sitemap.xml and declare it in robots.txt.
`robots_missing`	notice	robots.txt file missing	Publish /robots.txt with the Sitemap line.
`robots_blocks_audit`	error	robots.txt blocks the audit crawler	Allow the OrganiqAudit agent in robots.txt so the audit can crawl.

Which checks enter the site health score?

Errors alone count, and just the 17 page-level ones: a page with any of them counts as a page with an error. The 22 warnings, the 8 notices and the 5 site-wide checks, even the error among them, are listed and stay out of the calculation. No check can be switched off. The two formulas are in How a site health score is calculated.

How do you see the pages that fail a check?

Ask your assistant to call get_audit_issue with the code of the check, for example http_4xx. It lists the affected pages ordered by Search Console impressions. get_site_audit gives the issue counts of the whole audit. With site: "demo" both answer with fictional data and no account. Running the audit on your own site is described in SEO site audit in Claude and ChatGPT.

What does this checklist not cover?

It covers what the tables show and nothing beyond them. The crawl reads HTML without running JavaScript, so content a script adds is not seen. Core Web Vitals are read by another tool, get_core_web_vitals. Backlinks are not checked. No check promises a position: a fix removes a technical issue, and its effect is read later in Search Console.

Who is this checklist for, and who is it not for?

For whom: someone with a site verified in Search Console who wants a reference for each issue the audit reports and its fix. Not for whom: someone whose pages are written by JavaScript, who needs more than 48 checks or who audits sites that are not theirs; Screaming Frog or Sitebulb are the alternatives, with the exceptions in the next section.

How does this list compare with Screaming Frog and Sitebulb?

Read on November 10, 2026 from each vendor's own pages. On size, Organiq is behind: Screaming Frog and Sitebulb declare more than 300 checks against 48.

Factor: a second health score weighted by real Search Console impressions and an issue list ordered by the impressions of the affected pages, with the checks of this page behind both.
Comparison and exception: on the pages read on November 10, 2026, none describes a score weighted by impressions or a list ordered by impressions (Screaming Frog issues page, Sitebulb Audit Score and Hints, Ahrefs help article, Semrush Site Audit); the absence is on the pages read, not proof that nothing exists. Exceptions: all four run JavaScript and Organiq reads HTML; Screaming Frog and Sitebulb declare more than 300 checks against 48; their URL limits are larger; they crawl any site, with no verified property; Screaming Frog and Sitebulb label each issue by priority, while Organiq shows three severities; Screaming Frog adds GA4, command line and scheduling at any interval. Parity: Screaming Frog publishes its list too, on the issues page read on November 10, 2026; publishing the list is not a difference.
Where to see it: the tables above: each row has the code that get_audit_issue accepts, its severity group and its fix.

How was this page made?

Every row, code, severity, title, fix and count is generated from the list of checks in the code that runs the audit; a test checks that each code appears here. On November 10, 2026 we opened the vendor pages linked above. Not read in full: the Sitebulb Hints page, seen in a search summary. Corrections with a source: youssefpmat@gmail.com.

Frequently asked questions

What does a site audit check?

The Organiq site audit runs 48 checks: 18 errors, 22 warnings and 8 notices, on each page and on the whole site. This page lists every code with what it checks and how to fix it.

What is on a technical SEO audit checklist?

In Organiq the 17 page errors are HTTP status, titles, canonicals, sitemap, redirects, hreflang, mixed content and JSON-LD syntax; the warnings and notices cover descriptions, headings, images, links, speed and markup; 5 checks look at robots.txt, the sitemap, llms.txt and AI crawlers.

Which audit checks lower the site health score?

The 17 page-level errors, and no other check. Warnings, notices and the 5 site-wide checks are listed and stay out of the calculation.

How many checks does the Organiq site audit have?

48. Screaming Frog and Sitebulb declare more than 300, read on November 10, 2026.

How do I see which pages fail one check?

Ask your assistant to call get_audit_issue with the code of the check. With site "demo" it answers with fictional data and no account.