Sign inJoin the waitlist

Semrush audit: what Site Audit checks, how to run it and what to fix first

Over 140 checks sorted into errors, warnings and notices, with a fix order and a way to confirm the work.

Published by AI Knows Us · Updated 13 October 2026

The Semrush audit most people mean is Site Audit, the crawler inside Semrush that scans your website and reports technical and on-page problems. Semrush's own knowledge base page, read on 13 October 2026, says it checks for over 140 issues and sorts every finding into errors, warnings and notices. Run it, fix the errors first, re-crawl, and use the comparison view to confirm the count is falling.

What Site Audit checks

Semrush groups the checks into categories. The descriptions below follow its knowledge base page on what issues Site Audit can identify.

CategoryWhat it looks at
Crawlability and architectureSitemap, robots.txt, internal links, URL structure. Semrush says it covers over 20 common issues here.
On-page SEOTitles, h1 tags, meta descriptions, images and duplicate content
ContentDuplication, length and readability
Technical SEOHTML size, page speed, mobile optimisation, outdated technology
HTTPSCertificates, redirects, canonicals and mixed content after a move to HTTPS
International SEOHreflang language and country codes, conflicts and incorrect links
PerformanceLoading speed and the weight of HTML, CSS and JavaScript
Internal linkingLink distribution, incoming internal links, pages that need outgoing links, broken links
AI searchChecks aimed at AI-powered search, covered below

The same page lists AMP checks and says Site Audit covers over 40 AMP-related issues. Only sites that use AMP need to look at that group.

Errors, warnings and notices

Semrush orders findings by harm. Errors are the most harmful, warnings are harmful and notices are the least harmful. A sensible reading of that order is to work through errors completely, then the warnings that touch your most valuable pages, and treat notices as a tidy-up list. Typical errors include pages returning a 5XX server status, pages without title tags, duplicate title tags and hreflang conflicts. The knowledge base explains each issue with a "why it matters" paragraph and a "how to fix it" paragraph, which makes it a useful reference to keep open beside the report.

Run your first audit

  1. Add the site as a project. Use the exact version of the domain that serves your pages, with or without www, matching your canonical address.
  2. Set the crawl scope. Include the sections that matter and leave out areas that create endless URL variations, such as filtered search results.
  3. Run the crawl and open the overview. It shows the site health score, the issue counts by severity and the widgets that lead into each category.
  4. Open the issues tab and filter by errors. Export the affected URLs for each issue.
  5. Hand each issue to the person who can fix it. Content problems go to the editor, redirect and server issues to the developer.
  6. Re-crawl after the fixes. Compare the new crawl with the old one to confirm the errors have gone.

Semrush has a page on how to compare crawls and progress, which is the tool for step six. Use it after every batch of fixes so you can show what changed.

The order that fixes the most with the least work

  1. Server errors and broken pages. A page that returns an error cannot rank. Fix or redirect these first.
  2. Indexing blockers. Pages blocked by robots rules or tagged noindex by mistake, and broken canonical tags.
  3. Titles and headings. Missing or duplicate titles on your valuable pages.
  4. Internal links. Broken links, orphan pages and important pages with few links pointing to them. Our guide to internal linking goes through the audit.
  5. Speed. Heavy pages and slow loads, using the fixes in our Core Web Vitals guide.
  6. The notices. Batch them into a quarterly tidy-up.

Fix the pages that earn or could earn traffic before the pages that nobody visits. A crawl reports every URL equally, and your priorities should not.

The AI search checks

Semrush's page lists a category for AI search, also called generative engine optimisation. It describes checks for missing or incorrect llms.txt files (which it notes have been proposed as the AI equivalent of robots.txt), pages with overly long content that AI models might truncate, outdated last-modified headers that signal freshness, and pages with low semantic HTML use. These are readiness checks on your own site. They show whether the pages are in a form that machines can read well. They do not show whether an assistant names your business when a buyer asks, which is a separate measurement. Our page on Semrush AI visibility covers the product that tracks mentions.

How Site Audit fits with the rest of Semrush

  • The crawl finds problems on your own site. For the links pointing to it, use the reports in our guide to Semrush backlinks and the backlink audit.
  • For the wider set of reports in the order that makes them useful, see Semrush SEO analysis.
  • For plan limits that affect how many pages you can crawl, check the published plans in Semrush pricing and the 7 day offer described in the Semrush trial plan.
  • If you want to see what a free tool reports first, Semrush offers a free SEO checker page, and our overview of SEO audit tools compares the kinds of tool.

Reading the score without being led by it

The site health score is a summary of the issues found. A site can raise the score by clearing hundreds of notices on pages with no traffic and gain nothing in search. Treat the score as a progress bar for the work, and judge success by Search Console: impressions, clicks and the number of indexed pages for the sections you fixed. Record both before you start so the comparison is honest.

Questions about the Semrush audit

How many issues does Semrush Site Audit check?

Semrush's knowledge base states it can check over 140 website issues, with some checks combined to ease routine use.

Is Site Audit free?

Semrush offers a free SEO checker page and a 7 day free trial of its toolkits, at the time of reading. Check the current plans on the Semrush site before you rely on a limit.

How often should I run it?

Run it after every significant site change and on a monthly schedule for a site that changes weekly. A quarterly run is enough for a site that rarely changes.

What should I fix first?

Server errors, indexing blockers and missing titles on valuable pages. Those problems stop pages from being found or understood.

Does fixing every issue improve rankings?

Fixing errors removes obstacles. Rankings also depend on content, links and competition, so pair the audit with work on the pages themselves.

Readiness is one question; being named is another

A clean crawl tells you the site is readable. A buyer asking an assistant for a supplier in your category sends a different test, which is whether your business is among the names returned and whether the description is right. Run the audit, then ask ChatGPT, Gemini and Perplexity the questions a buyer would ask and compare the answers with the pages you just fixed. AI Knows Us (aiknowsus.com) takes that second step, recording for a business which assistants name it and which pages and outside sources they used.

Read next

Semrush SEO analysis: the five reports, in the order that makes them usefulSite Audit, Organic Research, Keyword Gap, Position Tracking and backlinks run in sequence, with a clear line between the numbers Semrush models and the numbers Search Console measures.AI Visibility Audit Deliverables for a UK Agency: The Five Artefacts, and Why the Zero Column Is One of ThemWhat a British agency should hand a client at the end of an audit, including the findings that did not survive a second reading.A Workflow Standard for AI Search Optimisation: Six Stations, and Who Owns Each OneThe execution sequence an Indian agency can hold a team to, measured against what our own site serves eleven named crawlers.The Multi Engine AI Citation Audit Framework: One Column Per Engine, Never a Blended ScoreA reproducible audit design for an Indian agency, built from 1,260 readings on 16 October 2026 and the 597 that came eleven days before it.A Small UK Agency Against a Large Consultancy on AI Search Optimisation: What the Citation Data Says Each One Can OwnThree British agency domains took dozens of citations each in our readings, and none of them is a large firm. Here is what that result does and does not tell a ten person shop.Build or Buy an AI Visibility Stack: The Five Silent Wrong Answers an In House Build ProducesWhat it actually takes to run a multi engine tracker yourself, measured from the defects in our own build, with rupee prices for the alternative.

Or browse the whole archive.

See what AI says about you.

The first scan is free, it needs no account, and it asks the questions your buyers ask.

Free. No card. We ask 5 real buyer questions on 2 AI apps.

Already sure you want this? Join the waitlist. We open on 31 October 2026. It costs an email address. No card, and nothing is charged. Or talk to a person if you have a question first. Or read how we fix what it finds.