How to run a full website SEO audit

What an audit is for

An SEO audit answers one question: what stops search engines from finding, understanding and showing your pages? It is not a score to chase. A score is only a summary. What you need is a short list of real faults, sorted so the ones that cost you the most get fixed first.

Most faults fall into a few groups. Google can’t reach a page. Google reaches it but is told not to list it. The page is listed but describes itself badly, loads slowly, or says too little to be chosen. An audit walks through these groups in that order, because a fault early in the chain hides everything after it. A beautifully written page that Google is blocked from reading will never rank.

Before you start: get the right access

You can learn a lot from the outside, but some answers only Google can give. Set these up first:

• Google Search Console for the website. It shows which pages Google has listed, which it left out and why, and the searches people used to find you. Our guide to reading Search Console data explains the reports.
• Your analytics tool, to see which pages bring visitors and sales, so you know which pages matter most.
• A login to the website or its hosting, so you can change robots.txt, the sitemap, redirects and page settings.
• A short list of your most important pages: the home page, the main product or service pages, contact and pricing. These get checked first and most carefully.

Step 1: crawl the whole site

A crawl is a program that starts at your home page and follows every link, the way a search engine does. It gives you the list of pages to audit and, just as useful, it shows the pages a search engine can’t find by following links.

1. Run a crawler on your home page. Free options include Screaming Frog SEO Spider (its free version crawls up to 500 addresses) and the page lists in Search Console. Larger sites need a paid crawler or several smaller crawls.
2. Export the list of addresses with their status codes. Every page should answer 200 (OK). Note anything that answers 301/302 (a redirect), 404 (not found) or 5xx (server error).
3. Compare the crawl with your sitemap. Pages in the sitemap that the crawl never reached have no links pointing to them. Pages the crawl found that are not in the sitemap may be missing from it, or may be pages that should not exist at all.

Step 2: can search engines reach the site?

Start with the faults that affect every page at once. They are rare, but when one is there, nothing else in the audit matters until it is fixed.

• The site opens on https://, and the http:// address sends visitors there.
• The site opens on one host name only. If both www.example.com and example.com show the site without one redirecting to the other, Google sees two copies of every page. Seoga’s scan reports this as “Your site opens on both www and non-www addresses”.
• The home page answers 200, not an error or a redirect loop.
• robots.txt exists, has no mistakes, and does not block pages you want found. The sitemap exists and opens. Both are covered in fixing robots.txt and the sitemap.

Step 3: can each important page be listed?

Reaching a page is not the same as listing it. A page can be reachable and still tell Google to leave it out, or point Google to a different page as the one to show. For each important page, check three things: it answers 200, it has no noindex tag, and its canonical tag (the “preferred address”) points to itself. Then check the Page indexing report in Search Console, which lists every page Google left out, with the reason.

Our crawl and indexing check goes through this step in detail, including what each reason in that report means.

Step 4: does each page describe itself well?

Search engines read a few parts of every page closely. Check them on your important pages first, then across the site:

• The title. Each page has its own, it says what the page is about, and it is short enough not to be cut off in results. Duplicate titles across many pages are a common finding.
• The search description. It does not change rankings directly, but Google often shows it under the title, so it affects whether people click.
• One main heading that matches the page’s topic, with sub-headings that follow in order.
• Picture descriptions (alt text) on images that carry meaning.
• Enough useful text. Seoga’s scan flags pages with almost no words as “pages have very little text”, and pages that are copies of each other as “pages are exact copies of another page”. Thin and duplicate pages are rarely chosen for results.

Step 5: speed and phones

Google ranks the phone version of your pages, so check on a phone-sized screen. Open your key pages in PageSpeed Insights, which shows Google’s Core Web Vitals for real visitors when it has enough data, and a lab test when it does not. Our Core Web Vitals guide explains the three numbers and the usual fixes.

Step 6: links inside your site

Links are how both visitors and search engines move through your site. From your crawl, look for links that lead to missing pages, links that go through a redirect when they could point straight at the final page, pages that no link points to, and pages buried many clicks from the home page. Each one is a small fault on its own; together they decide how easily Google finds and values your pages.

Step 7: sort the list by impact

A real audit of a real site finds dozens of things. Fixing them in the order you found them wastes time. Sort them by how much they block your pages, and how many pages they affect. Seoga’s scans use four levels, and you can use the same ones by hand:

Within each level, fix what affects the most pages, or your most important pages, first. A fault in a template (the shared layout every product page uses) is one fix that repairs hundreds of pages.

Step 8: fix, then check again

Change one thing at a time, and check the page again after each change. That way, if something goes wrong, you know what caused it. After a fix to an important page, you can ask Google to look at it again with the URL Inspection tool in Search Console (“Request indexing”). Google decides when it re-reads a page; that can take from a few days to a few weeks.