Every technical SEO audit starts with the same move: get a complete picture of what actually exists on your site. You can walk links manually on a ten-page brochure site, but on anything larger you need a crawler that follows internal links and sitemaps and records what it finds.

A good crawl captures far more than a URL list. For every page you want the final status code, the redirect chain it took, response time, canonical tag, robots meta directives, title, meta description, headings, word count, and every internal and external link. This raw inventory is the evidence base for every later step — without it, you are guessing.

If you would rather not set up tooling, DeepSEO Auditor does this automatically: enter your homepage URL and it crawls up to 500 pages, respecting robots.txt, and scores everything it finds.

Before optimizing anything, confirm search engines can actually reach your pages. Open robots.txt at your root domain — it must return 200, contain no accidental broad Disallow rules, and ideally declare your sitemap. Then check your XML sitemap: every URL in it should resolve without redirects, and every important page on the site should appear in it.

Next, look at indexation signals per page. Each important page should return status 200 directly (no redirect chains), be self-canonical or deliberately pointing at one preferred version, and carry no stray noindex tags left over from staging. Sitemap entries that redirect or are noindexed waste crawl budget and send mixed signals.

Finally, look at site structure: key pages should sit within about three clicks of the homepage, with no broken internal links and no orphan pages that exist in the sitemap but are linked from nowhere.

Metadata is the cheapest SEO win on most sites. Work through every indexable page and check the title tag first: unique, descriptive, and roughly 30–60 characters so it is not truncated in results. Duplicate titles are especially damaging — when two pages claim the same title, search engines must guess which one to rank.

Meta descriptions do not affect rankings directly, but a missing one invites Google to improvise your search snippet. Aim for 70–160 characters that sell the click. Then check headings: exactly one H1 per page, H2s under the H1, H3s under their H2 — no skipping levels for styling reasons.

On content itself, flag thin pages (under roughly 300 words) that compete for meaningful queries, and watch for near-duplicate content across templates. Keyword stuffing is rare by accident, but unnatural repetition reads as spam to users and search engines alike.

Internal links are how authority flows through your site and how crawlers discover pages. Check that important pages receive links from relevant pages — not only from the sitemap — and that anchor text describes the destination instead of generic phrases like "click here".

Then chase the mechanical problems: internal links pointing at redirects (each one is an unnecessary hop), broken internal links, and URLs that are very long or parameter-heavy. Faceted and tracked URLs should be canonicalized or disallowed so they do not create duplicate content.

If your site paginates archives, make sure each paginated page is linked and crawlable rather than reachable only through infinite scroll.

Structured data controls how your pages appear in rich results. Start simple: every JSON-LD block must parse without errors — malformed markup is silently ignored. Then confirm the right schema types exist for each page kind: Organization and WebSite sitewide, Article on blog posts, Product on product pages, BreadcrumbList wherever a breadcrumb trail shows.

Having a schema type is not enough: a Product block without price or availability will not earn rich results. Check that required and recommended properties are filled and that you do not have duplicate, conflicting blocks describing the same entity.

Round out the social layer: complete Open Graph tags (title, description, image) and Twitter card tags so shared links render a proper preview everywhere.

These checks decide trust. The entire site should be served over HTTPS with HTTP redirecting, and HTTPS pages must not load HTTP subresources — browsers flag or block mixed content. Pick one canonical host (www or non-www) and 301 the other to it, and handle trailing slashes consistently so /page and /page/ never both return 200.

Look at server response times next: slow time-to-first-byte hurts crawling efficiency and user experience, and pages that respond slowly are usually the same pages users abandon. Finally, confirm the viewport meta tag is present so pages render correctly on mobile, and that no production page accidentally carries a staging noindex.

A technical audit is only useful once it turns into an ordered to-do list. Sort findings by severity first: a site-wide noindex or robots.txt blocking your pages outweighs a handful of long titles. Within a severity level, prioritize by scope (how many pages are affected) and effort (a one-line fix beats a template rewrite).

Fix in waves — critical indexing problems first, then metadata and content, then link hygiene and structured data — and re-crawl after each wave. Comparing two runs is the only reliable way to know your fixes landed and nothing regressed.

A recurring audit cadence (monthly for most sites, weekly for large e-commerce) keeps new template changes and content publishes from quietly reintroducing old problems.