SEO checker: test your page on 15 points

Ok means a 200 response with no redirect, or after exactly one permanent redirect (301 or 308). More than one redirect, or a temporary one (302, 303, 307), is rated “Needs work.” According to Google’s documentation on redirects, only a permanent redirect signals that the target is the URL that counts. If a sixth redirect follows the first 5, the checker stops and reports an error. In your internal links, point straight to the final address.

Without HTTPS you get “Error.” “Needs work” means the page itself is encrypted, but images, scripts, stylesheets, videos or iframes are still loaded with http:// in the source code. The checker tells you how many. Change those addresses to https:// in the source.

If the checker finds “noindex” or “none” in the robots or googlebot meta tag or in the X-Robots-Tag header, you get “Error,” and it tells you where it found it. That’s only correct if the page really shouldn’t appear in search. Google only sees a noindex if robots.txt allows the page to be crawled. More on this under indexing and crawling.

The checker applies the rules the way Google interprets robots.txt: the longest matching rule wins, and in a tie the allowing one does. If the page is blocked, it shows the rule that applies. If the file is missing (a 4xx response other than 429), crawling is allowed and the point is Ok. If the server answers with 5xx or not at all, you get “Needs work,” because Google then stops crawling the whole site for the first 12 hours. A 429 (“too many requests”) is also rated “Needs work,” because Google doesn’t treat it as a missing file. A block in robots.txt prevents crawling. It does not remove a page from the index.

Ok is a canonical that points to the page itself, meaning the final URL after any redirects. Google recommends the self-reference. The checker reads the link in the HTML and the HTTP header. “Needs work” appears when the canonical is missing or points to another URL. In the second case you’re asking Google to show the other address, so check whether that’s intentional. Two different canonicals are an error because they contradict each other.

If the checker finds no hreflang links in the HTML, the point is Ok. It doesn’t see annotations in the HTTP header or the XML sitemap. If there are hreflang links in the HTML, it checks the spelling of the language codes and whether the page lists itself. Each language version has to list itself and all the others. The checker can’t see whether the other versions link back, because it fetches only one URL. If the return link is missing, Google ignores the annotations.

A missing or empty title is an error. “Needs work” appears with more than one title tag, with more than 60 characters and with fewer than 20. Both numbers are the checker’s rules of thumb. Google states no length limit and truncates the title in the search result to fit the width of the device. At 63 characters you don’t need to change anything, as long as the important part comes first.

If the description is missing, you get “Error.” That is a strict rating, because Google builds the snippet mainly from the page content and sometimes uses the meta description when it describes the page better. Without one, you give up that option. “Needs work” appears with more than one description, with more than 160 characters and with fewer than 70. These are rules of thumb too. Google has no fixed limit.

Without an H1 you get “Needs work.” Several H1s are rated Ok. If an H2 is followed directly by an H4, you also get “Needs work,” and the checker names the first jump. The reason is accessibility: assistive tools such as screen readers use headings to navigate within the page, and the W3C advises against skipped levels. To Google, the order makes no difference. This point turns yellow at most, never red.

Only images with no alt attribute at all are counted. An empty alt="" is not a defect, because an empty alt is correct for decorative images: screen readers then skip them. The checker can’t judge whether an existing alt text describes the image well. This point turns yellow at most, never red.

Ok means the checker found readable JSON-LD or Microdata. For JSON-LD it lists the types, such as Product or Organization. For Microdata it only reports that it is there. A JSON-LD block the checker can’t read as valid JSON is rated an error. Google accepts JSON-LD, Microdata and RDFa; the checker doesn’t detect RDFa. Google’s Rich Results Test shows whether required properties are missing and whether a rich result is possible. My page on structured data covers which types are worth the effort.

Without a viewport declaration, a phone shows the page at desktop width, scaled down. The checker rates that as an error. “Needs work” appears when the declaration is there but width=device-width is missing. The checker doesn’t test whether text and buttons are actually easy to use on a phone.

A value up to 2.5 seconds is good. Above 4 seconds you get “Error.” The largest element is, for example, the main image or a block of text visible without scrolling. The source follows the value in parentheses: field data, origin field data or lab measurement.

A value up to 200 milliseconds is good. Above 500 milliseconds you get “Error.” INP needs interactions; simply loading a page doesn’t measure it. That’s why the checker reads INP from field data only. If there is no INP value there, you get “No data.”

A value up to 0.1 is good. Above 0.25 you get “Error.” CLS rises when images without fixed dimensions, late-loading banners or fonts push the content someone is reading or about to tap out of place.