If a new website is not indexed after a week, inspect the specific URL in Search Console before deciding what to fix. One week is not a Google deadline or evidence of a penalty. An accidental exclusion or access error needs repair. A crawled-but-unindexed page also needs a review of its current content, duplicate relationships and usefulness, rather than repeated submission.
1. Establish the URL and the version you are inspecting
Copy the final public address, including HTTPS, hostname, path and any meaningful parameters. Open the matching Search Console property and enter the complete URL in URL Inspection. Save the indexed status, last crawl time, crawl and indexing permissions, user-declared canonical and Google-selected canonical. Then run a live test and inspect the returned HTML, screenshot and resources.
The URL Inspection documentation distinguishes indexed information from a live test. A repaired page passing the live test does not mean the indexed record has updated, and the test does not cover every indexing condition. If the address redirects, inspect the final target separately instead of treating the old URL's indexed result as the target's result.
Absence from a site: search alone is insufficient to confirm that a page is unindexed. Without Search Console access, you can still inspect responses, content and links, but leave the index status unverified rather than inventing a definite cause.
2. Let the reported state guide the next check
| Observation | Check first | Next action |
|---|---|---|
| Unknown URL or incomplete discovery information | Public navigation, sitemap and intended address | Add genuinely relevant discovery paths |
| Discovered, currently not indexed | No crawl yet; availability and unnecessary URL generation | Repair demonstrated problems and observe subsequent crawling |
| Crawled, currently not indexed | Crawled revision, complete content, duplicates and canonicals | Address a specific gap; do not translate the status into plagiarism or a penalty |
| Blocked by robots.txt | Rules for the correct host, protocol, port and path | Remove only unintended blocks on public content |
| noindex detected | HTML and response headers; whether exclusion is intentional | Remove accidental exclusions on intended index targets |
| Duplicate or alternate page | Whether the chosen canonical is appropriate | Keep a valid relationship; correct an inappropriate target |
| Server, redirect or soft-404 problem | Responses, every redirect hop and actual body content | Restore the appropriate status and content, then check later GSC observations |
The Page indexing report documentation defines “discovered” as known but not yet crawled. A crawled-but-unindexed page may or may not be indexed later; that status does not require repeated resubmission. Neither label proves a failure to meet a word count, backlink quota or AI-detector score.
3. Inspect more than HTTP 200
Examine a full GET response, not merely whether a browser opens the page. This screenshot records an actually executed local teaching server. It is not a Google crawl, Search Console screen or client website.

| Local request | Observed response | What the check establishes |
|---|---|---|
| /ok | 200; no noindex in this example HTML or headers | Absence of these signals, not proof of indexing |
| /meta-noindex | 200; noindex in HTML | Accessibility and indexing permission differ |
| /header-noindex | 200; X-Robots-Tag: noindex | An HTML-only search would miss this control |
| /old | 301; Location: /ok; followed response 200 | Inspect the old address, redirect and target separately |
| /duplicate | 200; canonical points to /ok | A declaration was found; Google acceptance was not simulated |
| /gone | 404 | The local server explicitly reported not found |
| /blocked and /robots.txt | Page 200; robots 200 with Disallow: /blocked | An ordinary request can work while a crawler rule disallows the path; Googlebot was not simulated |
Google's HTTP status guidance says a 200 response does not guarantee indexing. An error message or empty content returned with 200 may be treated as a soft 404. Repair routing, access or rendering faults on pages that should exist. A genuinely removed page without a replacement does not need an unrelated homepage redirect just to make an error list smaller.
4. Separate robots, noindex and canonical handling
robots.txt controls crawling; noindex controls indexing
Inspect the public host's root robots.txt and the rule matching the intended URL. Google generally caches robots for up to 24 hours, but inability to refresh can extend that period; 24 hours is not a guaranteed update deadline. In the robots.txt specification, 4xx responses for the robots file are generally treated as no valid crawl restrictions, except 429. A 403 for robots.txt and a 403 for an article are different cases.
The robots-file 5xx behaviour has stages and cache conditions. For the first 12 hours Google stops crawling the site while retrying robots. For the next 30 days it tries to use the last good version; without a cached version it assumes no crawl restrictions. If the problem persists beyond 30 days and the site is otherwise available, Google acts as though no robots file exists while checking for a new version. If the site also has general availability problems, Google stops crawling the site but still periodically requests robots. These rules describe the robots file, not a universal countdown for every page returning 503. Fix the failure promptly rather than treating those windows as safe downtime.
Google must be able to fetch a page to see a noindex meta tag or response header. A robots-disallowed URL can still appear through information found elsewhere; a crawl block is not a reliable substitute for an indexing exclusion. Use access controls for private material, and distinguish deliberate protection from an accidental block on public content.
A canonical is a preference signal, not a visibility switch
Compare the current page, declared canonical and Google-selected canonical. Are they duplicates or very similar content? A product page accidentally pointing to the homepage needs correction. A tracking-parameter copy pointing to the same product's clean URL may be appropriate. Do not send every parameter page to the homepage, or canonicalize a translation to another language simply because the subject matches.
The canonicalization guide recommends consistent redirects, canonical annotations, sitemaps and internal links. Google still evaluates the canonical choice. Reading a tag establishes the signal you supplied, not that Google accepted it.
5. Inspect discovery paths and the content actually returned
Follow a relevant category or article link to the target. Use parseable a href links with descriptive anchors. JavaScript-generated links are not automatically uncrawlable; examine the rendered links and content. Include preferred canonical URLs in the sitemap, while remembering that sitemap submission is a discovery hint rather than a guarantee that each URL will be crawled or indexed.
Read the fetched or tested body. Is it a login screen, challenge, unavailable-product message or empty application shell? A complete page in your desktop browser does not establish what every access environment receives. Once the intended content is available, check whether it fully answers the question, uses verifiable evidence and adds something beyond near-duplicate pages on your own site. An originality percentage cannot substitute for that review.
Keep the repair and Google's later response separate
- Record the affected URL, demonstrated cause, old response and content, and modification time.
- After deployment, verify public responses, current content and essential resources. Confirm that intentional exclusions remain intact.
- Request indexing where appropriate for a few important URLs you manage, or maintain a sitemap for larger sets. Repeated requests for the same URL do not accelerate crawling.
- Check later crawl dates and indexed information. Record “live fault repaired,” “recrawled” and “indexed” as separate observations.
For an indexed page without useful query visibility, continue with the content and search-intent investigation. For ongoing work, use the first-month plan. The Chinese version covers the same troubleshooting sequence.