Technical SEO
Technical SEO Audit Checklist: 60+ Checks, Including AI Crawlers
A technical audit is only useful if it finds the problem that's actually holding the site back. This is the checklist I work through on every client site, in the order that finds the big problems first.

A technical SEO audit checks whether search engines and AI crawlers can discover, crawl, render, index and understand every important page on a site, and finds what's stopping them. A good audit is prioritized: it looks for site-wide indexing blockers first, then errors that affect many pages, then warnings. It uses a crawler, Google Search Console and PageSpeed Insights together, and it ends with fixes and a re-crawl, not just a report.
I've run this checklist on lead-generation sites, e-commerce stores, large content sites and my own site after it was hacked. The results range from routine to dramatic: a real estate site went from 84% to 100% Semrush Site Health in 16 days (case study), an e-commerce store cleared 787 errors and warnings (case study), and a UK car dealer turned out to be blocked from Google entirely (case study).
Tools you need
- A crawler: Semrush Site Audit, Screaming Frog or similar, to see the site the way a bot does.
- Google Search Console: Pages report, URL Inspection, Sitemaps, Core Web Vitals, Removals, Links, Settings.
- Bing Webmaster Tools: crawl errors and index coverage for Bing, which matters for several AI assistants.
- PageSpeed Insights: lab and field performance data.
- Rich Results Test and Schema Markup Validator for structured data.
- Server or CDN logs, when available, to see real crawler behavior.
Phase 1: critical blockers (check these first)
These can make a whole site invisible. I check them in the first hour.
- [ ] Site-wide noindex. Check the homepage and a few key pages for
<meta name="robots" content="noindex">and theX-Robots-Tagheader. WordPress's "Discourage search engines" setting causes this more often than you'd think. - [ ] robots.txt blocking. Read the whole file. Look for
Disallow: /, blocked CSS and JavaScript, and broad AI-crawler blocks. My guide to AI crawlers and robots.txt lists the bots that matter. - [ ] Canonical tags pointing elsewhere. A site-wide canonical to the wrong domain or the homepage tells Google to ignore everything else.
- [ ] HTTPS and host consistency. One canonical version (https, with or without www), with 301 redirects from the others.
- [ ] Server errors. 5XX responses on important pages, or a firewall that blocks Googlebot.
- [ ] Manual actions and security issues in Search Console.
- [ ] Search Console Pages report. How many pages are indexed versus excluded, and why.
The UK car dealer audit is the classic example: the homepage returned "This site is blocked from indexing". Until that setting was removed, no content or keyword work could bring in traffic, so it was priority one.
Phase 2: crawling and indexing
- [ ] XML sitemap exists, lists only indexable 200-status canonical URLs, and is submitted in Search Console and Bing Webmaster Tools.
- [ ] Sitemap isn't set to noindex. In one e-commerce re-audit, the sitemap itself was set to noindex. Product, page and HTML sitemaps were rebuilt and resubmitted.
- [ ] Sitemap listed in robots.txt with a
Sitemap:line. - [ ] Excluded pages reviewed. For each reason in the Pages report, check whether important pages are affected. "Crawled - currently not indexed" deserves its own investigation; see crawled, currently not indexed.
- [ ] Orphan pages. Pages in the sitemap with no internal links pointing to them.
- [ ] Crawl depth. Important pages reachable within three clicks from the homepage.
- [ ] Parameter and faceted URLs under control, so crawlers don't waste time on endless filter combinations.
- [ ] Duplicate content. Near-identical pages consolidated or canonicalized.
- [ ] JavaScript rendering. Main content present in the server-rendered HTML, or at least rendered reliably. Compare the page source with Search Console's rendered HTML.
Phase 3: status codes and redirects
- [ ] Broken internal links (4XX). Fix the link or redirect the target.
- [ ] Broken external links and images. In one real estate audit, 12 broken external images were replaced or removed.
- [ ] Redirect chains and loops. Every redirect should go to the final URL in one step.
- [ ] 302 redirects that should be 301. Temporary redirects for permanent moves send mixed signals.
- [ ] Soft 404s. Pages that say "not found" but return 200.
- [ ] Removed content. Return 404 or 410 for pages that are gone for good. For a hacked site, I return 410 for spam URLs so they drop out faster; the full process is in removing hacked spam pages from Google.
- [ ] Old URLs with backlinks redirected to the closest relevant page instead of returning errors.
Phase 4: on-page technical elements
- [ ] Title tags: present, unique, descriptive and not truncated. The real estate audit found 7 overlong titles; a NYC studio had 29.
- [ ] Meta descriptions: present and unique. The NYC studio had 58 pages without one.
- [ ] H1: exactly one per page, describing the page.
- [ ] Heading hierarchy: no skipped levels used for styling.
- [ ] Internal links wrongly set to nofollow. The e-commerce store had 141 internal links set to nofollow; the real estate site had 10.
- [ ] Thin internal linking. The NYC studio had 45 pages with only one internal link pointing to them.
- [ ] Image alt text on meaningful images.
- [ ] Language and hreflang tags correct for multilingual or multi-country sites.
- [ ] Mixed content: HTTPS pages loading HTTP resources or linking to HTTP pages.
Phase 5: speed and Core Web Vitals
- [ ] Core Web Vitals report in Search Console: URLs rated Poor or Needs improvement.
- [ ] PageSpeed Insights on key templates, mobile first.
- [ ] Largest Contentful Paint: main image or heading appears quickly.
- [ ] Interaction to Next Paint: pages respond quickly to taps and clicks.
- [ ] Cumulative Layout Shift: nothing jumps while loading.
- [ ] Unminified or render-blocking CSS and JavaScript. The real estate site had 21 unminified files; the e-commerce re-audit found 450.
- [ ] Image formats and sizes, caching and server response time.
The details and fixes are in the Core Web Vitals guide. For reference, one travel booking site went from 42 to 97 on mobile in two days of focused speed work (case study).
Phase 6: structured data
- [ ] Organization or Person markup on the homepage with
sameAslinks. - [ ] BreadcrumbList on inner pages, matching visible breadcrumbs.
- [ ] Article, Product, Service or LocalBusiness where relevant.
- [ ] No duplicate or conflicting blocks from plugins and themes.
- [ ] Markup matches visible content.
- [ ] Valid JSON: one syntax error breaks the whole block.
My recommendations for which markup to use are in schema markup for AI search.
Phase 7: AI search readiness
This phase didn't exist in audits a few years ago. Now it's standard on every project.
- [ ] AI search crawlers allowed: OAI-SearchBot, PerplexityBot, Claude-SearchBot, plus Googlebot and Bingbot.
- [ ] CDN and firewall not blocking or challenging those crawlers.
- [ ] Snippet controls: no unintended
nosnippet, restrictivemax-snippetordata-nosnippeton key content, since Google applies these to AI features. - [ ] Bing Webmaster Tools verified, sitemap submitted, no crawl errors.
- [ ] Answer-first structure on key pages, with question headings and direct answers.
- [ ] llms.txt present and valid if you choose to use one. It's optional, but Lighthouse now checks its format.
Phase 8: security and access
- [ ] Search Console users and owners. In one e-commerce audit, an unrecognized user with access to the store's Search Console was found and removed. Most audits never check this.
- [ ] Unexpected pages in the index. Search
site:yourdomain.comand look for spam, foreign-language or casino pages, signs of a hack. - [ ] Malware and injected scripts. Check Search Console's Security issues report and scan the site.
- [ ] Outdated CMS, plugins and themes, the most common entry point for spam injection.
- [ ] Backlink profile spikes. A sudden wave of toxic domains can signal spam attacks; see the toxic backlinks guide.
Phase 9: migrations, domains and international sites
Run these checks whenever a site has moved, merged or targets several countries.
- [ ] Old domain redirects. Every old URL redirects to its closest new equivalent with a 301, not all to the homepage.
- [ ] Change of Address submitted in Search Console after a domain move.
- [ ] DNS pointing to the right server. After a hosting move, stale DNS can serve an old, broken or hacked copy of the site to some visitors and crawlers. I've had to fix exactly this on my own site.
- [ ] hreflang return links. Every language or country version references the others, and each references itself.
- [ ] Country targeting consistency. Currency, addresses, phone formats and content match the market each version targets.
WordPress-specific checks
Many of the sites I audit run on WordPress, and it has its own recurring problems:
- "Discourage search engines from indexing this site" left on after launch, which adds a site-wide noindex.
- Several SEO plugins at once, each adding its own titles, canonicals and schema, often contradicting each other.
- Plugin-generated sitemaps that include tag pages, attachment pages and author archives you don't want indexed.
- Attachment pages for every uploaded image, creating hundreds of thin URLs.
- Page builders such as Elementor adding heavy CSS and JavaScript that slow down mobile pages.
- Outdated plugins and themes, the most common way spam pages get injected.
In Rank Math or Yoast, I check the sitemap settings, noindex rules for archives and the schema output on every audit. Getting these settings right often clears dozens of warnings at once.
Prioritizing what you find
A full crawl can report hundreds of issues. I sort every finding into three tiers:

| Tier | What it includes | When to fix |
|---|---|---|
| Critical | Indexing blocks, site-wide noindex, robots.txt blocking, server errors, hacks | Immediately |
| High | Broken links, redirect chains, missing titles and descriptions, nofollow internal links, poor mobile Core Web Vitals | Within two weeks |
| Medium and low | Overlong titles, missing alt text, minor warnings | During regular maintenance |
The aim isn't a perfect score for its own sake. It's removing everything that stops important pages from being crawled, indexed and trusted.
What a useful audit report looks like
I've reviewed plenty of audits from other providers, and the weak ones share a pattern: a 60-page PDF exported from a tool, with every warning listed and no priorities. A useful report is shorter and sharper:
- A one-page summary: the three to five issues that matter most, in plain language, with their business impact.
- A prioritized issue list: each issue with the affected URLs, why it matters, how to fix it and who should fix it.
- Baseline numbers: Site Health, indexed pages, Core Web Vitals status, so improvement can be measured.
- Evidence: screenshots from Search Console and the crawler.
- A re-crawl after fixes showing the before and after.
Red flags in cheap audits
- No Search Console access requested. An audit without Search Console misses indexing data, manual actions and security issues.
- Only a score. "Your SEO score is 67" tells you nothing about what to fix first.
- Everything marked urgent. If every item is critical, nothing is prioritized.
- No mention of the business. A good audit connects technical issues to the pages that make money.
Fix at the source, then re-crawl
A report that lists problems without fixing them changes nothing. My process after the audit:
- Fix at the source: in the CMS, theme, server config or templates, so the issue doesn't reappear on the next crawl.
- Validate in Search Console: use URL Inspection's live test on key pages, resubmit sitemaps and request indexing.
- Re-crawl: run the same crawler again and compare before and after, issue by issue.
- Monitor: re-audit monthly or quarterly, because new problems appear with every plugin update and content change.
Frequently asked questions
How often should I run a technical SEO audit?
Run a full audit when you start working on a site, after any redesign or migration, and at least once or twice a year. Lighter monthly checks of Search Console and a scheduled crawl catch new problems, such as a noindex sitemap or broken links, before they cost traffic.
What is a good Semrush Site Health score?
There's no official threshold, but scores in the 90s usually mean no serious technical problems remain. The score matters less than what's behind it: one site-wide indexing block is worse than dozens of minor warnings. Prioritize errors that affect indexing and important pages.
Can I do a technical SEO audit myself?
You can cover the basics with Google Search Console, a free crawler and PageSpeed Insights by working through a checklist like this one. Complex issues such as JavaScript rendering, large-scale duplication, migrations and hacks usually benefit from someone who has fixed them before.
How long does it take to fix technical SEO issues?
It depends on the site and the issues. For many small and medium sites, the main problems can be fixed in two to four weeks. In one real estate project, Site Health went from 84% to 100% in 16 days. Rankings then respond as Google recrawls the fixed pages.
Does technical SEO affect AI search visibility?
Yes. AI answer engines can only cite pages they can crawl and read. Blocked crawlers, missing indexing, JavaScript-only content and snippet restrictions can all keep a page out of AI answers, no matter how good the content is.