- Home
- Technical and performance
- HTTP Status Checker
HTTP Status Checker
Check what your server actually answers, independently of what your browser shows.
A direct HTTP request from a server, with no cache and no cookie. The time measured is that of the first byte returned. No usage limit.
What this tool is for
Your browser lies, unintentionally. It keeps pages in cache, carries your cookies, remembers your previous visits. A page can display perfectly there while returning an error to a first-time visitor or to a crawler.
This tool queries the server the way a stranger would: no cache, no cookie, no history. The code returned is the one Google and your new visitors actually receive.
It is the first check to make when a page does not index, when a link seems dead, or when a visitor reports a problem you cannot reproduce. In most cases the answer is in the HTTP code.
The reflex that saves time: before imagining a Google penalty, check the response code. A page returning 403 to crawlers will never be indexed, and no content optimisation will change that.
How to read the results
200 is the normal answer: the page exists and is served.
301 and 302 signal a redirect. The address requested is not the one served. The redirect checker then sets out the whole journey.
403 means access is refused. On your own site that is often an application firewall blocking automated requests — and so blocking Googlebot too. It is a cause of complete deindexing, invisible while browsing normally.
404 indicates a page that does not exist. Normal for a mistyped address, problematic if links still point at it.
500 reveals an execution error on the server. Your hosting error logs will give the precise cause.
503 announces temporary unavailability. That is the code to use during maintenance: Google comes back later without deindexing, provided it does not last for weeks.
The response time reads separately. Below 200 milliseconds, excellent. Past one second, your server is the first brake on loading.
Common mistakes
Testing from a browser with a cache. The most common case: the page displays for you and nowhere else. Always test under neutral conditions.
Ignoring a 403 thinking it only concerns bots. Googlebot is a bot. If it gets a 403, your page does not exist for Google.
Using a 404 during maintenance. Google then reads the page as deleted and starts deindexing. A 503 with a Retry-After header is the right answer.
Settling for a single test after an error. An error can be transient. Run several spaced-out measurements before concluding there is a structural problem — or that there is none.
Checking only the home page. It is often served from a cache and hides problems that only affect dynamic pages.
Going further
- Redirect CheckerFollow the full redirect chain of a URL, hop by hop
- SSL Certificate CheckerValidity, chain of trust and expiry date of a certificate
- Core Web Vitals CheckerReal field data from users
Frequently asked questions
My site displays but the tool returns an error, why?
Because your browser uses its cache and your cookies, while the tool arrives as an anonymous visitor. A server can also refuse automated requests with a 403 while answering browsers normally. Test in a private window to get closer to the conditions of a first-time visitor.
What does a 503 code mean?
That the service is temporarily unavailable: maintenance, overload, or a restart. It is the only error code Google treats gently — it comes back later without deindexing. It is therefore the right code to return during maintenance, with a Retry-After header.
What response time should I aim for?
Below 200 milliseconds for the first byte is excellent. Up to 600 is acceptable. Past one second the server becomes the main brake on loading, and no image optimisation will make up for it. That is often the sign of a saturated shared host or a badly indexed database.
Why test from several countries?
Because a server can be reachable from one country and blocked elsewhere, or dramatically slower. That is frequent when a firewall filters by geographic zone, or when no delivery network is in place. If you target several markets, this check is indispensable.
Is a slow site penalised by Google?
Server response time is not a direct factor, but it governs the LCP, which is one. A server taking a second to answer makes a good LCP nearly impossible, whatever you optimise downstream. So it is an indirect but decisive factor.
What if my server answers intermittently?
Run several spaced-out measurements: an isolated error can be a passing incident. If the intermittence is confirmed, it often comes from saturation at peak hours or a resource limit on shared hosting. Your hosting error logs will give the exact cause.