How to run a technical SEO audit, step by step

A technical audit answers one question: can search engines reach, read and trust your pages? Work in this order, because each step makes the next one meaningful.

1. Check the front door

Before crawling anything, load these four addresses by hand:

• http://yoursite.com should redirect to https://yoursite.com
• https://www.yoursite.com and https://yoursite.com should settle on one of the two, not serve both
• yoursite.com/robots.txt should exist and not disallow everything
• yoursite.com/sitemap.xml should list real pages

Two versions of a site serving the same pages splits your links between them. One redirect fixes it.

2. Crawl the site

Crawl with a tool that follows links the way a search engine does, and note:

• 4xx errors. Dead pages that other pages still link to. Either restore them, redirect them to the closest replacement, or remove the links.
• 5xx errors. Server failures. Repeated 5xx responses make crawlers slow down and eventually drop pages.
• Redirect chains. A link to page A that redirects to B, which redirects to C. Point the link straight at C.
• Pages that need JavaScript. If the HTML arrives nearly empty, crawlers see an empty page.

3. Look at titles and descriptions across the whole site

Single pages are easy to check by eye. Problems appear at scale:

• Titles used on more than one page, usually from a template.
• Titles over 60 characters, cut off in results exactly where the useful part sits.
• Missing descriptions, where the engine invents a snippet from whatever text it finds first.

4. Follow the internal links

Three numbers tell you most of what you need:

• Pages nothing links to. They exist, but only the sitemap knows about them.
• Click depth. Pages more than three clicks from the homepage get crawled less.
• Most linked pages. This is what your own site says is important. If your money pages are not near the top of that list, your navigation disagrees with your business.

5. Measure speed where it counts

Two different things get called speed:

• Lab tests (Lighthouse, PageSpeed Insights) run one simulated visit. Useful for finding causes: oversized images, blocking scripts, slow servers.
• Field data (Chrome UX Report) reports what real visitors experienced over the last 28 days. This is what Google uses.

A lab score of 95 with poor field data usually means real visitors are on slower devices or connections than your test.

6. Check what structured data you publish

Structured data does not lift rankings on its own, but it lets search engines show rich results and helps AI assistants understand what a page is. At minimum: Organization on the homepage, and the type that matches each page (Article, Product, FAQPage).

7. Check the AI crawlers separately

Your robots.txt may allow Googlebot and block the crawlers behind AI answers, or a firewall may block them without you knowing. That is now its own audit, covered in the AI search readiness guide.

Priorities, when everything is broken at once

1. Anything that blocks indexing: noindex, robots.txt rules, canonical tags pointing at the wrong page.
2. Errors on pages that matter: 4xx and 5xx on money pages and their links.
3. Duplicate titles and descriptions on important pages.
4. Speed problems that affect real visitors.
5. Everything else.

Rankwell's site audit runs all of the above in one pass: it crawls the site, checks each page against more than 30 rules, works out the real click depth and inbound links from the link graph, and scores AI readiness separately. Each finding comes with why it matters and what to change, so the report is a work list rather than a wall of numbers.