Free technical SEO audit checklist

A template that 500s wastes the crawl.

Check the homepage, a service, a category, and a detail page.

HTML at that URL is usually a soft 404.

Open /robots.txt. It should be plain text, not the homepage.

A stray disallow is the usual accident.

Test the path. Read the line that won.

Google needs them to render.

Do not disallow /wp-includes or the theme assets.

Every sort and filter can become an indexable URL.

Disallow the parameter patterns you do not want crawled, and canonical the rest to the clean URL.

Blocking the whole admin folder can block a file the front end uses.

Disallow /wp-admin/ and allow /wp-admin/admin-ajax.php.

www and non-www both returning 200 splits the site.

Pick one. Redirect the other. Use it in canonicals and the sitemap.

A chain through www and a trailing slash is three hops.

Make the first redirect the final URL.

Tag archives and thank-you pages are the usual yes. Services are not.

Search the theme and the SEO plugin for noindex.

Excluded is a reason, not a score.

Open the samples. Fix the reason, not the count.

A thin page that says nothing gets dropped.

Give it content or return a real 404.

Printer pages, parameters, and http copies are the usual set.

One URL should answer for each piece of content.

Page 20 of a tag is rarely a landing page.

Noindex them or fold them, and keep the links crawlable.

A missing canonical lets parameters win.

The tag should be the clean URL of that page.

A 302 on a permanent move looks temporary.

Use 301 for URLs that have moved for good.

Three hops is a habit, not a plan.

Point the old URL at the final URL.

A loop burns the crawl and shows an error.

Fetch the URL and follow until it stops.

Redirecting every dead URL to the homepage is a soft mess.

Redirect only when there is a real replacement.

Links to redirected URLs keep the chain alive.

Update the menu, the content, and the sitemap.

A relative canonical is easier to get wrong.

Use the full https URL.

A canonical to the other host fights the redirect.

One host, everywhere.

Sort and currency parameters should not be the indexed URL.

The clean URL is the canonical, and it returns 200.

Noindex on a URL that others canonical to is confusing.

Pick one instruction.

en-uk is wrong. en-GB is the region.

Use ISO language and region codes.

A one-way hreflang is ignored.

Each alternate should list the others, including itself.

Pointing hreflang at a redirected URL fails the cluster.

Use the final URL.

It is the fallback, not a duplicate of every country.

Point it at the page a person should see when no region matches.

HTML here is a broken route.

Open it. You should see urlset or sitemapindex.

Noindex URLs in the sitemap contradict the page.

Remove redirects, 404s, and noindex URLs.

One giant file gets ignored past the limit.

Use a sitemap index. Keep each file within the published limits.

It is a hint. Still worth one line.

Add the Sitemap line.

Plugins and files overwrite each other.

Fetch the live URL and compare it with the setting.

Training and search are different bots.

Allow search if you want to be cited. Block training only if that is the decision.

Blocking Google-Extended does not remove you from Google Search.

Read the bot before you toggle it.

That directive has been unsupported since 2019.

Use a meta tag or an HTTP header.

Google ignores it.

If you need to slow a crawl, use Search Console or the host, not that line.

Client-only text is a coin flip for some crawlers.

View source. The H1 and the first paragraph should be there.

A div with a click handler is not a link.

Use an href.

A JavaScript app that returns 200 for missing URLs creates soft 404s.

The server status should match the page.

A canonical injected late can be missed.

Put it in the head from the server.

Lab data is a hint. Field data is what users hit.

Use PageSpeed or Search Console. Note when the sample is too small.

It is usually the hero image or the heading.

Do not lazy-load the hero. Give it a size.

Heavy scripts are the usual cause.

Cut the tags. Break long tasks.

Ads, fonts, and images without a size are the usual cause.

Set width and height. Reserve space.

One syntax error drops the block.

Validate it. Fix the line it names.

A price in schema that is not on the page is a problem.

Use the same name, price, and availability the person sees.

Two nodes with the same id overwrite each other.

Give each node its own id.

FAQ rich results ended for most sites on 7 May 2026.

Use it to mark up real questions. Do not sell the rich result.

A small site can live without them. A large one should not.

Ask the host for access logs, or use the platform export.

Parameter URLs eating the crawl show up here.

See which paths get the hits. Block or canonical the junk.

A weekend of 500s drops pages.

Watch the error rate after a deploy.

Mixed content breaks the lock.

Update hardcoded http links.

It tells browsers to stay on HTTPS.

Set it once the redirects are clean.

Open redirects get abused.

Do not redirect to any URL a query string names.

A hack warning kills clicks.

Check the security report. Clean the site before you chase rankings.

Blocking one is not the same as blocking the other.

Read the purpose, then toggle.

One trains, one searches, one fetches when a person asks.

Block the one you mean. Leave the search bot if you want citations.

It is the Gemini training opt-out.

Leave Googlebot allowed if you want to rank.

A clear paragraph is easier to cite than a slogan.

Write the answer in the HTML.

Adoption by the big answer engines is not confirmed.

Add the file if it is easy. Do not expect it to change visibility on its own.