Search
Enter at least two characters without personal information.

    LibraryArticles & FAQ Your projectRequest a quote
    LanguageEN
    Practical guides

    Diagnose a page missing from Google

    Publication does not guarantee indexing, and indexing does not guarantee visibility for a particular query. Start with the exact URL and observed state before rewriting the page.

    Go to the method

    Review of search results on a laptop
    AI-generated illustration

    Define what is actually missing

    Record the URL, publication date, search query and intended destination. In a Search Console property you are authorised to access, compare URL inspection with the live test. Inspection describes Google’s known state; the live test checks the page currently accessible.

    A site: query or title search alone is not a reliable index inventory. Separate four questions: does Google know the URL, can it crawl it, has it selected it for indexing and does it appear for useful searches? These require different checks.

    Scroll the table horizontally to read every column. With a keyboard, focus the table area and use the arrow keys.

    Read the result in its time context
    Situation or settingWhat it meansUseful check
    Indexed-version dataDescribes the version Google knew when it collected the page.Compare collection and fix dates before inferring failure.
    Live test succeededThe page appears accessible to the tool at test time.This guarantees neither indexing nor appearance for a particular query.
    Declared and selected canonicals differThe declared preference differs from Google’s known choice.Check content, redirects and links; the live test does not predict canonical selection.

    Check technical obstacles

    Open the URL without signing in, examine its HTTP status and redirect chain, then look for noindex in HTML and X-Robots-Tag in headers. Check robots.txt and server refusals. Test essential resources too if the content depends on JavaScript.

    robots.txt controls crawling; it is not access protection and does not guarantee removal of a URL from results. A crawler must be able to access a noindex instruction to read it. Private pages need access control. Clarify whether the page should be public before removing protection.

    Look for the instruction in the response that actually serves the content. An HTML page can deliver noindex in a meta tag; a PDF can deliver it through the X-Robots-Tag HTTP header without an HTML tag. A 200 response with noindex may therefore be an intentional exclusion. Google does not support a noindex rule written in robots.txt.

    For Google to read noindex, its crawler must be able to access the resource. Simultaneously blocking crawling can prevent that instruction from being read. After a correction, record the current test separately from the last known crawl date: a successful live test does not instantly update the index or predict the selected canonical. Confidential resources remain protected by authentication independently of these search signals.

    Scroll the table horizontally to read every column. With a keyboard, focus the table area and use the arrow keys.

    Indexing instructions: check the delivery method and access
    Observed situationWhat Google can readUseful check
    Accessible HTML with meta noindexThe page instruction when the crawler fetches it.Confirm that exclusion is intended before changing the tag.
    Accessible PDF with X-Robots-Tag noindexThe instruction in the file’s HTTP header.Check the PDF response itself, not just its linking page.
    Crawling blocked and noindex on the resourceThe block may prevent reading noindex.Compare publication intent, robots.txt and the response accessible to the crawler.
    Successful live test after correctionAn accessible state at the time of testing.Then compare the known crawl; do not infer guaranteed indexing.

    Check URL and content consistency

    Compare the declared canonical with Google’s selected URL where available. A canonical pointing elsewhere may explain exclusion of a variant. Menus, internal links, sitemaps and redirects should support the intended destination rather than signal contradictory versions.

    If the page is accessible and properly signalled, assess its distinct value: does it answer a real question, provide specific information and belong to a useful internal journey? Adding repetitive paragraphs to nearly identical pages does not solve insufficient coverage. Consolidate equivalents where that helps the reader and use a relevant destination.

    Follow the address chain before interpreting an inspection verdict. Fictional example: https://example.com/ancienne/ redirects over HTTP to https://example.com/nouvelle/. Google says live testing follows redirects before examining the page without necessarily indicating this or displaying the final tested URL. A positive result therefore does not prove that the old address itself serves the content. Record its status, destination header, intermediate steps and final response separately.

    The entered URL’s indexed state and live result do not always describe the same version or step in that chain. Also inspect the exact destination URL in an authorised property covering it. Compare crawl date, declared canonical and Google’s known choice when available. The live test does not predict canonical selection. Retain both observations instead of treating one green screen as validation of a redirect, indexing or ranking.

    Scroll the table horizontally to read every column. With a keyboard, focus the table area and use the arrow keys.

    Fictional redirect: entered address versus examined page
    ObservationScopeCheck to retain
    HTTP response from /ancienne/Starting addressRedirect status and stated destination
    Live test started on /ancienne/May examine /nouvelle/ after redirectCheck chain and final URL separately
    Indexed state of /ancienne/Known index state of entered addressDo not treat it as the destination’s indexed state
    Separate inspection of /nouvelle/Exact destination in a suitable propertyCompare crawl and canonical; no ranking guarantee

    Correct and observe without promising timing

    Retain the diagnosed problem, correction, date and successful checks. Update links and the sitemap if the URL changes. An indexing request can signal a correction but does not guarantee an inclusion date or ranking.

    Recheck a representative set of pages rather than republishing the whole site after every fluctuation. If exclusion is intentional, document that decision. The useful result is a verifiable action: restore access, improve, consolidate, retain outside the index or protect with authentication.

    Check Google controls and measure AI impressions

    Google documentation consulted on 9 October 2026 now distinguishes an inclusion control for generative Search features and a dedicated performance report. In Search Console, check the relevant property’s setting and any inheritance before interpreting missing visibility. Inclusion establishes eligibility, not a guaranteed citation. This control does not replace indexing, snippet controls or controls concerning model training.

    The generative AI report describes impressions in the covered features. Compare consistent periods, final pages, devices and countries. An absent report may reflect insufficient volume; it does not by itself establish exclusion. Separate impressions from visits and actual enquiries. Without property access, both the setting and results remain unverified.

    Content updated on

    Documented reference points for this topic

    Documented reference points for this topic
    ReferenceValue or syntaxScope
    Google Search · Canonical
    rel="canonical"A preference signal for Google; neither a redirect nor a selection guarantee.
    Google Search · Sitemap
    50 000 URL · 50 MBLimits per uncompressed file; larger sets require several sitemaps.

    Dates indicate when references were consulted. These facts are neither a site audit nor an overall compliance statement.

    Read an HTTP response before fixing the page

    Choose a situation to separate the status meaning, the useful check and the pitfall. This reader explains cases; it does not test any address.

    200Available page
    What it means
    The request succeeded. For an HTML page, also examine the returned content.
    What to check
    Check the title, main content and language, then the canonical and indexing directives.
    Watch out for
    A “not found” page returning 200 may become a soft 404. HTTP success does not guarantee indexing.
    301 / 308Permanent move
    What it means
    301 and 308 indicate a permanent new address. 308 explicitly preserves the method and body; with 301, a client may change POST to GET.
    What to check
    Open the old URL and check Location, a relevant destination and the absence of chains or loops.
    Watch out for
    Do not redirect every old page to the home page. For a form, also check the method expected at the destination.
    302 / 307Temporary move
    What it means
    302 and 307 indicate a temporary change. 307 preserves the method and body; with 302, a client may change POST to GET.
    What to check
    Check that the destination is available and that a return to the original address is planned.
    Watch out for
    A temporary redirect does not describe permanent removal. Do not replay a payment or order without checking its state.
    404 / 410Missing or removed resource
    What it means
    404 does not say whether the absence is temporary or permanent. 410 indicates a likely permanent removal.
    What to check
    Look for a broken link, a changed URL or deliberate removal. Offer a replacement only when it meets the same need.
    Watch out for
    A helpful error page must retain the actual 404 or 410 status. An error message with 200 does not fix the absence.
    429Too many requests
    What it means
    429 indicates request rate limiting. A Retry-After header may say when to try again.
    What to check
    Examine the limit’s scope and Retry-After. Slow down requests and bound retries.
    Watch out for
    Avoid immediate retry loops. Without an HTTP response, a network failure is not a 429 status.
    503Temporarily unavailable service
    What it means
    503 fits temporary unavailability, such as maintenance or overload. Retry-After may provide a delay or date.
    What to check
    Check the actual status, affected scope and return to normal service. Remove the notice after recovery.
    Watch out for
    Do not replace all pages with an error indefinitely. Persistent 5xx responses can eventually cause Google to remove URLs from its index.

    Technical references : RFC 9110 §15 · RFC 6585 §4 · Google · HTTP.

    Acceptance matrix to adapt to your project

    These proposed checks use synthetic cases. Decide the required behaviour with the team, record the result and assign unresolved gaps before release.

    Scroll the table horizontally to read every column. With a keyboard, focus the table area and use the arrow keys.

    Test cases, expected outcomes and useful evidence
    CaseExpected outcomeEvidence to retain
    Public accessThe intended page works without login or error.Exact URL, received status and content rendered without a session; live-test date.
    DirectivesCrawling and indexing match the intention.robots.txt, meta robots and headers.
    CanonicalThe destination agrees with links and sitemap.Declared URL and observed selection.
    CorrectionAn identified problem has a verifiable response.Date, test and state follow-up.

    Frequently asked questions

    Will every sitemap URL be indexed?

    No. A sitemap helps discovery of proposed URLs. Crawling and indexing remain subject to technical checks and search-engine selection.

    Should robots.txt be removed to fix the problem?

    First check the relevant rule and whether the page should be public. Deleting the whole file does not fix noindex, server errors or conflicting canonicals and may open unnecessary crawl paths.

    Should an indexed page with low visibility be rebuilt?

    Start with intended queries, answer relevance and observed performance. Indexing and ranking are different issues; low visibility does not automatically justify a redesign.