All checks

The blue headline in search results comes from here; without it Google makes one up.

A short title takes up little space in search results and carries few keywords.

It is cut off in search results, so the important part should come first. 61–70 characters is informational, above 70 is a warning.

Google chooses which one to use on its own.

Pages with the same title compete with each other, and Google cannot tell which one to show.

The description text in search results is picked at random from the page.

It does not fill the space in search results and can lower the click-through rate.

Info: the description ends with an unfinished phrase such as "…in which." or "…and"; it looks incomplete in search results, so the text should be finished.

It is cut off in search results.

Google chooses which one to use on its own.

It makes it harder to tell pages apart in search results.

The heading that states the page's main topic is missing.

Usually not a problem; the main topic should be in a single H1.

A heading tag with nothing in it; most likely a template error.

The template adds the same long text to every title, so titles are cut off in search results. One setting shortens it. Pages that are long even without the addition still get 'title too long' separately.

Info: jumps such as H2 to H4 break the content structure; screen readers and bots lose the hierarchy.

Info: if the description repeats the title and always ends with the same closing, it loses what is unique to each page.

Google will not show this page. Not a problem if it is intentional.

The page is noindex but its canonical points to another address that returns 200 and can be indexed; this is a common deliberate setup for duplicate and filter pages. If it is not intentional, noindex and canonical together are redundant.

On parameterized or duplicate addresses it is harder for Google to choose the main page.

This page is left out of the index and the address it points to is indexed instead. An error if it is not intentional.

Info: addresses with parameters such as ?sort= are tied to the main page by canonical; this is the correct setup. Search checks such as title and description were not run on these addresses.

With conflicting canonicals Google ignores all of them.

The canonical points to a page that does not return 200 or is noindex.

Google ignores a language code it does not recognize.

There should be a single address per language.

Every language version should list itself too.

The version for visitors whose language matches none is not specified.

The language version it points to does not return 200 or is noindex.

If version B does not point back to A, Google does not pair the two.

The language version's title is the same as another language version's or was left in English; the page matches poorly for searches in that language.

The description is the same as another language version's or was left in English; foreign-language text shows in search results.

The page states its own language differently in two places; Google cannot know which one to trust.

The page links to an address that returns 4xx/5xx.

Every redirect uses up crawl budget; the link should point to the final address.

It is in the sitemap but no page links to it; Google treats it as unimportant.

Most pages link to a broken target; it comes from the template, such as the menu, header or footer. It is fixed in one place, and the 'broken internal link' findings on all pages go away together.

Most pages link to an address that redirects (menu, header or footer). It is fixed in one place by changing the link to the final address.

It receives a link from only one page; more internal links help it rank.

Info: a nofollow link to your own pages passes no value. Links marked sponsored/ugc (intentional) and external links are not counted.

Pages far from the home page are crawled less often and receive less value.

Info: texts such as 'click here' or 'read more' do not tell Google what the target page is about.

An indexable page is missing from the sitemap; Google may take longer to find it.

The sitemap should contain only addresses that return 200 and can be indexed.

The sitemap should list the final address, not the one that redirects.

Conflicting signal: the sitemap says 'index it', the page says 'do not index'.

The sitemap should contain only canonical addresses.

The sitemap declared in robots.txt returns 404/5xx; Google cannot discover pages this way.

A broken sitemap cannot be read by Google, or is read only in part.

A single sitemap can hold at most 50,000 addresses and 50 MB; the excess is ignored, so it should be split into a sitemap index.

Info: Google does not trust a lastmod that does not give real modification dates and ignores it.

An http address in the sitemap when the site is https causes redirect and duplicate problems.

A sitemap may contain only addresses of its own domain; the others are ignored.

Google drops a page that returns errors from its index; if it keeps happening, it also lowers the crawl rate.

Visitors and Google land on an empty page, and the value of every link pointing to this address is wasted.

Not a problem on its own, but internal links and the sitemap should point straight to the final address.

A timeout or connection error; the server may be slow or unreachable.

The server returns a success code but the page has nothing in it; Google may treat this as a 'soft 404'. The page should either exist or return 404. Missing title/description/H1 for the same page is not reported separately.

In image search and for screen readers it is unclear what the image shows. (alt="" counts as decorative and is not a finding.)

The page can shift when the image loads (CLS); the browser cannot reserve space. Images whose space is reserved with an overlay class (absolute inset-0) or aspect-* are not counted.

Modern formats are noticeably smaller at the same quality, so the page opens faster. The server or image pipeline should produce WebP/AVIF.

It slows down page loading and uses mobile data.

It seriously reduces page speed (LCP); it should be compressed or made smaller.

Info: total and largest sizes of the site's unique images, measured with HEAD requests.

If loading=lazy is set on an image at the top of the screen, the largest content (LCP) loads late.

Structured data that cannot be parsed is ignored entirely by Google.

Without @context https://schema.org the markup is not recognized as a schema.

An entity with no type is not used, because it is unclear which rich result it qualifies for.

Values such as lorem ipsum / TODO / example.com were left over from the template; they publish wrong information.

url/image/logo fields must be absolute addresses; a relative address fails validation.

Info: these types no longer give rich results on Google. Harmless; there is no need to remove them.

The title/image of a link shared on social networks and messengers comes out broken or random.

Shares and likes are recorded against the wrong address; usually a mistake left over from the template.

Info: no large-image card appears when shared on X/Twitter.

If robots.txt returns 5xx, Google may stop crawling the site.

Not required, but crawl rules and the sitemap cannot be declared.

robots.txt could not be reached; the rules could not be checked.

robots.txt blocks the root directory; the site will not appear on Google. A development setting may have been left in place.

This line is the easiest way for search engines to find the sitemap.

Info: these bots cannot read your site, so you appear less in AI answers. It may be a deliberate choice.

Google cannot render the page visually; mobile friendliness and content may be judged wrongly.

Every hop slows things down and uses crawl budget; link value can shrink with each hop.

The page never opens; visitors and Google see an error.

The same content is live at two addresses; ranking signals are split and visitors stay on the insecure version.

Visitors and Google reach the home page through several hops; dropping from https to http causes a security warning and loses link value.

The domain you entered redirects to another site; there are no or very few pages to crawl on this domain.

The domain you entered opens, but its pages declare the www or non-www version as canonical; the crawl was run on the domain Google will index.

The first visit is needlessly slow; a single 301 is enough.

The site is live on two domains at once; risk of duplicate content and split signals.

On mobile the page opens at desktop width; Google indexes mobile-first.

Googlebot processes only the first 2 MB of HTML; content and links after that are not seen.

Info: a page with no content sends visitors to another address with JavaScript (location.replace). A server-side 301/302 is more reliable for bots and link value.

If content arrives only through JavaScript, bots and AI crawlers see an empty page. Reported only when the body is just an empty app root or has no text at all; short pages with real text get 'thin content' instead.

Site-wide: median and 90th percentile response time. It is a pattern, not a momentary fluctuation on one page; slow pages are re-measured before judging.

Scripts in <head> without async/defer delay the page from appearing (LCP).

A slow server lowers both the user experience and Google's crawl rate.

Long addresses are hard to share and read, and are cut off in search results.

/Page and /page can count as different addresses; risk of duplicate content.

Google prefers hyphens as word separators; underscores join words together.

Parameterized addresses can duplicate the same content; the canonical should point to the main address.

The same page opens separately at /a and /a/; one should redirect to the other.

The server shows every file in the folder with an 'Index of /…' page; backups, configuration or private files may leak. Turn directory listing off (Options -Indexes) or put an index page in the folder.

The same http:// resource (an image, ad or embedded map in the template) is loaded on many https pages. It is fixed in one place, and the 'mixed content' findings on those pages go away together.

An https page loads an http:// resource; the browser may block it or show a 'not secure' warning.

The browser does not learn to always connect to the site over https; the first visit can fall back to http.

The browser may guess the file type; 'nosniff' turns this off.

The full page address (including parameters) can be passed on when visitors go to other sites.

Info: an extra layer of protection against XSS attacks is missing. It does not directly affect SEO.

The server returned 429. This is not a problem with your site but with the crawler; these pages could not be checked.

Info: part of the site was not crawled; the numbers in the report do not reflect the whole site.

Pages with little text rank poorly. On helper pages such as contact, privacy, tags and cart it is reported as info (being short is natural); another deliberately short page can be marked 'intentional'.