Guides · 5 min read · Updated 2026-09-08
The technical SEO audit checklist (2026): 40 checks in the order that matters
A working technical SEO audit checklist, ordered by impact: crawlability, indexability, duplication, links, content, speed and structured data — with what to look for and how to fix each item.
Most technical SEO checklists are lists of everything a crawler can measure. This one is ordered by what actually moves rankings and revenue when it’s wrong, because an audit’s job is to produce a prioritised list of changes a developer will make, not a spreadsheet of 400 rows. Work top to bottom; stop when you run out of budget, knowing the important things are done.
Before you crawl: three questions
- What is the canonical hostname?
https://www.example.comorhttps://example.com, never both. Every other variant should 301 to it. Check all four combinations of scheme and www. - What should not be indexed? Staging, internal search results, filtered category pages, account pages. Write it down; you’ll compare the list against what the crawl finds indexable.
- How big is the site? Count the URLs in the XML sitemap and compare with
site:in Google. A large gap either way is your first finding.
Section 1: Can search engines reach the pages?
Robots.txt. Fetch it. Confirm it doesn’t block CSS, JavaScript or image directories, and that it references the sitemap. A Disallow: / left over from staging is still the most expensive single line in SEO.
Response codes. Every page in the sitemap should return 200. Anything else is a finding: 404s waste link equity, 5xx pages are dropped from the index after repeated failures, and 429s mean the server is rate-limiting crawlers, which throttles Googlebot too.
Redirect chains and loops. Each hop loses time and, past a handful, Googlebot gives up. Chains usually come from stacking migrations: http→https→www→trailing-slash. Point every source directly at the final URL.
Internal links to redirects and errors. The redirect itself is fine; the link that goes through it isn’t. Fix the link, not just the redirect.
Orphan pages. Pages in the sitemap that nothing links to. Google can find them, but with no internal links they rank as if nobody cared. Either link them or drop them from the sitemap.
Crawl depth. Pages more than four clicks from the home page get crawled less often. Flatten the architecture with hub pages and pagination that isn’t a hundred pages deep.
Section 2: Should these pages be indexed, and are they?
Noindex on pages that should rank. A single stray noindex on a template can remove a whole category. The crawl’s Directives tab shows every noindexed page; compare with your list from before.
Indexable pages that shouldn’t be. Internal search results, filtered and sorted listing variants, parameter URLs. These dilute the crawl budget and create duplicates. Noindex them or canonicalise them to the base listing.
Canonical tags. Every indexable page should have exactly one, pointing at itself or the preferred version. Missing canonicals on parameterised pages, canonicals to non-indexable pages, and canonicals that disagree with the sitemap are the three common failures.
Sitemap accuracy. The sitemap should contain indexable, 200, self-canonical pages and nothing else. A sitemap full of redirects and noindexed URLs tells Google the sitemap can’t be trusted.
Hreflang. For multilingual sites: every alternate must link back, codes must be valid BCP-47, and every page must include itself. One missing return link invalidates the pair.
Section 3: Is the content unique and substantial?
Exact duplicates. Usually URL variants: trailing slashes, uppercase paths, tracking parameters, print versions. Canonicalise and redirect.
Near duplicates. Product pages that differ only by colour, location pages with the same template text, paginated archives. Decide which one should rank and canonicalise the rest, or make them genuinely different.
Thin pages. Under about 200 words of main content with no other purpose. Tag pages, empty categories, placeholder posts. Expand, merge or noindex.
Soft 404s. Pages that say “no results” or “not found” but return 200. Return a real 404 or 410.
Titles and descriptions. Missing, duplicated across pages, over the pixel limit, or the same as the H1 verbatim. Titles are still the strongest on-page signal; duplicated titles are the most common finding on every audit.
Headings. One H1 per page that describes the page. Skipped levels and multiple H1s are minor; a missing H1 on a template is a real finding.
Section 4: Do the links work for it?
Anchor text. “Click here” and “read more” tell search engines nothing. Internal links should describe their target.
Link equity to important pages. Count internal links per page (Truelint calls this inlinks and computes a Link Score). Money pages with three inlinks and blog posts with three hundred is a common inversion.
External links. Broken outbound links are a user problem; links to pages that now redirect to spam are a trust problem. Re-check them annually.
Nofollow on internal links. Almost never right. It leaks equity into nowhere.
Section 5: Rendering and speed
JavaScript rendering. Crawl once with rendering on and once off, and diff. If titles, canonicals or links only exist after JavaScript runs, some search engines will miss them and Google will see them late. Server-render anything that matters.
Mobile. Content wider than the viewport, text under 12px, tap targets under 48px. Test at 375px, which is where most of your users are.
Page weight. Images without dimensions, uncompressed images, render-blocking scripts. Core Web Vitals live in PageSpeed Insights; a crawler can flag the causes.
Security. HTTPS everywhere with no mixed content, HSTS, a content security policy, and no forms posting to HTTP.
Section 6: Structured data
Validity. JSON-LD that doesn’t parse is ignored silently. Validate every page type.
Required properties. A Product without offers, an Article without datePublished, a Breadcrumb without position: each disqualifies the rich result.
Consistency. Schema that disagrees with the visible page (a price in the markup that isn’t on the page) is a manual-action risk.
Turning findings into a plan
Group findings by template, not by URL. Two thousand product pages with a missing canonical is one ticket. Estimate each ticket’s effort and its reach (how many pages, how much traffic), and sort by reach divided by effort. Send the top ten with the exact fix wording. Then re-crawl in a month and compare: the comparison, not the original audit, is what proves the work happened.
Doing this in minutes
Every item above is a check in Truelint, with the issue list already grouped and each finding carrying its fix. Crawl the site, read the Issues pane top to bottom, export the rules that matter, and re-crawl later to compare.