Technical SEO Audit Checklist for 2026
Every check we run in a technical SEO audit, in the order we run them, with what passing looks like and a free checklist sheet to track your fixes.
By The Serpzeno team · · 12 min read
Download the checklist (CSV)A technical SEO audit answers one question: can search engines find, crawl, render and index the pages you care about, and is anything on the site holding those pages back? This checklist walks through the checks in the order we run them, from the ones that can remove a page from Google entirely down to the ones that polish how it looks in results.
Each item says what to check and when it passes. You can work through it by hand with Google Search Console and a browser, or run an automated audit and use the list to read the results. There is also a downloadable checklist sheet (CSV, opens in Google Sheets or Excel) with a status column for tracking fixes.
Before you start: what you need
- Google Search Console access for the site. Its Pages report and URL Inspection tool show what Google actually did with your URLs, which no outside tool can see.
- A crawler that follows your internal links and, ideally, renders JavaScript. This is how you find broken links, orphan pages and duplicate titles across the whole site.
- PageSpeed Insights for Core Web Vitals: field data from real Chrome users where Google has enough of it, plus a lab test.
- A list of the pages that matter most: the ones that earn traffic, leads or sales. Fix problems on those first.
1. Crawling and robots.txt
If a crawler can't reach a page, nothing else on this list matters for it. Start with /robots.txt.
- robots.txt loads. It returns status 200 at the root of the domain (or a 404, which Google treats as "no rules"). A 5xx error makes Google treat the whole site as temporarily blocked.
- Important pages aren't disallowed. Test your key URLs against the rules for Googlebot. A stray
Disallow: /left over from a staging site is one of the most common causes of sudden drops. - It isn't used to hide pages from the index. Blocking a URL in robots.txt stops crawling, not indexing: a blocked page can still appear in results without a description. To keep a page out, allow crawling and use
noindex. - AI crawler rules are a decision, not an accident. GPTBot, ClaudeBot, Google-Extended and others read robots.txt too. Decide which you allow for training and which for search answers, and write it down.
- Your sitemap is listed. A
Sitemap:line with the full URL helps every crawler find it, not just the ones you submit to.
Our free robots.txt checker tests any page against Googlebot, Bingbot and ten AI crawlers and shows which rule decided each result.
2. XML sitemap
- It lists only indexable, canonical URLs. No redirects, 404s,
noindexpages or parameter duplicates. A sitemap full of URLs you don't want indexed teaches Google to trust it less. - It stays within limits. Each file holds at most 50,000 URLs and 50 MB uncompressed; bigger sites use a sitemap index.
- lastmod dates are real. Only set
<lastmod>when the content changed. Dates that update on every build are ignored. - It is submitted in Search Console. And the Sitemaps report shows it was read without errors.
The XML sitemap checker counts the URLs in any sitemap or sitemap index and flags relative URLs, other domains, http/https mixes, duplicates and invalid dates.
3. Indexability
Open Search Console's Pages report. It groups every URL Google knows about by why it is or isn't indexed, and it is the fastest way to see whether the site has a real problem.
- Important pages are indexed. Use URL Inspection on your top pages. "URL is on Google" is a pass.
- No accidental noindex. Look for
<meta name="robots" content="noindex">and theX-Robots-TagHTTP header on pages that should rank. CMS settings and plugins add these more often than people expect. - 'Crawled – currently not indexed' is understood. Google fetched these pages and chose not to index them, usually because they look thin or too similar to others. Improve or merge them rather than resubmitting.
- 'Discovered – currently not indexed' isn't growing. Google knows the URLs but hasn't crawled them yet. A large or growing number often means weak internal links to those pages or too many low-value URLs competing for crawl time.
- Soft 404s and server errors are fixed. Empty category pages, "no results" pages that return 200 and intermittent 5xx errors all appear here.
4. Canonicals and duplicate content
- Every indexable page has a self-referencing canonical.
<link rel="canonical">with the page's own absolute URL, matching the version you link to (https, with or without www, with or without a trailing slash). - Canonicals point at pages that return 200. A canonical to a redirect, a 404 or a
noindexpage sends Google mixed signals. - Google agrees with your choice. "Duplicate, Google chose different canonical than user" in the Pages report means it didn't. Look at what makes the pages look alike.
- Near-duplicate pages are merged or made distinct. Location pages, tag archives and product variants with the same text compete with each other.
5. Site structure and internal links
- No orphan pages. Every page you want indexed is linked from at least one other page. Pages that only exist in the sitemap get crawled less and rank worse.
- Important pages are close to the home page. Key pages should be reachable in about three clicks. Deeply buried pages are crawled less often.
- No broken internal links. Links to 404 pages waste crawls and lose the value the link would pass. Fix the link, don't just add a redirect.
- Links are real links. Use
<a href>elements. Navigation that only works through JavaScript click handlers isn't followed by crawlers. - Anchor text describes the target. "Rank tracking for agencies" tells Google more than "click here".
6. Status codes and redirects
- Old URLs redirect with a 301. Permanent moves use 301 (or 308), straight to the final URL.
- No redirect chains or loops. Keep each redirect to a single hop. Chains slow crawling and users, and loops make pages unreachable.
- http and non-canonical hostnames redirect.
http://, and www or non-www (whichever you don't use), should each redirect to the one canonical version in one step. - Deleted pages return 404 or 410. Not a redirect to the home page, which Google usually treats as a soft 404 anyway.
7. Titles, descriptions and headings
These are on-page items, but they belong in a technical audit because problems usually come from templates, not people.
- Every page has a unique title. Duplicate titles across many pages almost always mean a template is missing a variable.
- Titles fit. Google cuts titles by pixel width, roughly 600 px on desktop. Around 50–60 characters is a safe target.
- Meta descriptions exist and are unique. 120–160 characters that summarise the page. Google may rewrite them, but a good one is used more often.
- One clear H1 per page. It says what the page is about and usually matches the title's main phrase.
Check a single page with the meta tag checker, or preview how a title and description will look in Google with the SERP snippet preview.
8. Structured data
- The JSON-LD is valid. Invalid JSON means the whole block is ignored. Test key templates in Google's Rich Results Test.
- It describes what is on the page. Organization on the home page, Product on product pages, Article on posts, BreadcrumbList on deeper pages. Marking up content users can't see breaks Google's guidelines.
- Search Console's enhancement reports are clean. Errors there stop pages from getting rich results.
9. JavaScript rendering
Google renders JavaScript, but rendering happens after crawling and can fail or time out. If your content, links or meta tags only exist after scripts run, check that Google sees them.
- Main content is in the rendered HTML. Use URL Inspection → View crawled page, or compare the raw HTML with what a rendering crawler sees.
- Titles, canonicals and noindex aren't changed by scripts. If JavaScript rewrites them, Google may index the version before or after the change. Put the final values in the server response.
- Content doesn't need a click to load. Crawlers don't click "load more" buttons or open tabs that fetch content on demand.
10. Page speed and Core Web Vitals
Google's thresholds for a "good" experience, measured at the 75th percentile of real visits:
- LCP (Largest Contentful Paint): 2.5 seconds or less.
- INP (Interaction to Next Paint): 200 milliseconds or less.
- CLS (Cumulative Layout Shift): 0.1 or less.
- Field data passes on key templates. The Core Web Vitals report in Search Console groups URLs by template. Fix the worst group first.
- The LCP element loads early. Usually the hero image or heading. Don't lazy-load it, size it correctly and serve modern formats like WebP or AVIF.
- Nothing shifts as the page loads. Give images and embeds a width and height, and reserve space for banners and ads.
- Long tasks are kept short. Lab metrics like Total Blocking Time point at heavy JavaScript, the usual cause of a poor INP.
11. Mobile and HTTPS
- Pages have a viewport meta tag. Google indexes the mobile version of your pages, so the mobile view must contain the same content and links.
- The whole site is on HTTPS. A valid certificate, http redirecting to https and no mixed content (images or scripts loaded over http).
12. International sites (if you have them)
- hreflang tags are reciprocal. If the English page points to the German one, the German page must point back. One-way annotations are ignored.
- Codes are valid. Language codes like
enoren-GB, plusx-defaultfor the fallback page.
13. Images
- Meaningful images have alt text. It describes the image for screen readers and image search. Decorative images use an empty
alt="". - Images are compressed and sized. No 3000-pixel photos shown at 300 pixels. Modern formats are usually much smaller.
- Social share images are set. Open Graph tags (
og:title,og:image) control how links look when shared.
How to prioritise what you find
A full audit usually finds more than you can fix in a week. Work in this order:
- Anything that removes pages from Google: robots.txt blocks, stray noindex, server errors, broken canonicals.
- Problems on your money pages: the ones from your list of important pages, even if the issue looks small.
- Template-wide issues: one fix in a template can repair hundreds of titles or canonicals at once.
- Everything else, by how many pages it affects.
Re-run the audit after each round of fixes and watch the Pages report in Search Console. Indexing changes take days to weeks to show up.
Run the checklist automatically
Serpzeno runs 28 of these checks on every page it crawls, across on-page, content, technical, performance and security, and grades each area A to F. It also ranks the issues so you know what to fix first. See a real audit report to know what you get, or compare plans on the pricing page. The Free plan audits one website with no credit card.
For the checks that need Google's own data, like indexing status and real-user Core Web Vitals, connect Search Console: our Search Console reports turn it into a monthly summary. Google's SEO Starter Guide is a good companion if you're new to this.
Questions
How often should I run a technical SEO audit?
Run a full audit at least once a quarter, and after any redesign, migration, CMS change or big content release. Small problems like broken links appear all the time, so an automated weekly check is worth having.
What is the difference between a technical SEO audit and an on-page audit?
A technical audit checks whether search engines can crawl, render and index the site: robots.txt, sitemaps, status codes, canonicals, speed. An on-page audit checks the content of each page: titles, headings, copy and keywords. Titles and meta descriptions sit in both because problems usually come from templates.
Can I do a technical SEO audit for free?
Yes. Google Search Console and PageSpeed Insights are free, and this checklist covers what to look at. A crawler saves time on larger sites; Serpzeno has a Free plan that audits one website.
How long does a technical SEO audit take?
For a small site, a few hours by hand. Automated tools crawl hundreds of pages in minutes; the time goes into reading the results and deciding what to fix first.