A successfully submitted XML sitemap does not mean the pages inside it have been indexed. The sitemap file is a discovery input, not a page-by-page indexing receipt. Investigate three layers: whether Google can read the file, whether it can discover and fetch the intended pages, and how it treats those pages after processing.
1. Confirm that Google read the sitemap you intended
Open the correct property in the Sitemaps report. Check the exact submitted URL, latest status, last-read time and discovered-page count. A local XML file or a CMS “generated” message does not establish what the public server delivered.
If the sitemap could not be fetched, inspect the public response, access restrictions and the report's details. If it was read with errors, fix the reported structure or entries. Use the fetch, parsing and temporary-error guide for those cases. Repeatedly deleting and submitting an unchanged file does not identify the cause.
For a sitemap index, inspect its child sitemap files as well. The index lists file locations; its presence does not prove that every child is current or readable. Follow Google's creation guidance for intended canonical URLs, correct encoding and truthful lastmod values. Do not update every date merely because the generator ran again.
Once the status is Success, move to the page URLs. Google's documentation describes queueing parsed URLs but also says discovery in a sitemap does not guarantee crawling or indexing. The discovered-page count is not an indexed-page count.
2. Check how the important pages are discovered and fetched
Select representative URLs rather than diagnosing the whole site from one total: an important service or product, a recent article, an older missing page and a comparable indexed page. Record their type, sitemap membership, internal entry, response status, canonical, indexing rules and last observed Google fetch.
Google's link guidance explains crawlable links and descriptive anchors. Make important pages reachable through useful navigation, categories or relevant content. A scoped crawl and inventory comparison can identify missing entry paths, but the crawler's own discovery is not proof of a Google fetch.
Use current live tests and actual server/CDN logs to investigate access. Check the response body as well as status: a 200 page can still show an empty template, challenge or error. HTTP handling distinguishes normal content, redirects, most 4xx errors and the special 429/server-error case. Fix observed failures rather than adopting a universal response-time cutoff.
For JavaScript pages, compare the initial and rendered content using Google's rendering guidance. Essential content that only appears after user interaction needs a different implementation. For product templates, the six technical product-page checks provide a more detailed sequence.
3. Read the actual page-level indexing reason
The Page indexing report groups reasons for pages not being indexed. Combine it with indexed data in URL Inspection; a current live test is not an override of the indexed result and does not predict Google's final canonical choice.
| Observation | What it means for the investigation | Useful next step |
|---|---|---|
| Discovered — currently not indexed | Google knows the URL but has not crawled it | Review useful entry paths, server capacity, duplication and content priorities; no fetch deadline is implied |
| Crawled — currently not indexed | Google fetched it but has not indexed it | Inspect the fetched content and page value; the status alone does not identify one definitive cause |
| Duplicate / a different canonical was chosen | Another representative may be used | Compare the tested, declared and selected pages and their signals |
| Excluded by noindex | An exclusion rule was read | Keep it if intentional; otherwise fix the source and verify a later fetch |
| Blocked, error or soft 404 | Access or delivered-content problems need investigation | Reproduce the exact URL response and inspect the actual body |
| Indexed, but little search visibility | Indexing and query performance are different | Review intent, queries, impressions and clicks in equivalent periods |
The report reason is an investigation starting point, not a universal explanation for every URL in a group. “Crawled — currently not indexed” does not by itself prove a penalty. A duplicate can be an intended alias, and a noindex can be a deliberate exclusion. Fix the pages that should be independently available, rather than forcing all URLs into one target state.
Use URL-level evidence instead of treating site: as a census
Google's site: operator documentation warns that results are not exhaustive. A displayed count or missing result is not a precise inventory of indexed URLs. Search Console's known-URL reports and URL inspections answer more specific questions, with their own dates and scope.
Do not calculate an “indexing rate” by dividing a loose site: estimate by a sitemap count containing duplicates, excluded pages or a different time period. Define the exact intended URL set first. Also keep indexing, impressions, clicks, AI citations and qualified enquiries as separate outcomes.
After a repair, record the change and the observation
- Save the affected sample and the reason for the change.
- Verify the corrected public response, rendered content, rules, internal links and relevant sitemap entry.
- For a few important changed URLs, use the appropriate recrawl request; for many URLs, maintain the sitemap. Repeating the same request does not accelerate crawling.
- Watch subsequent fetches and indexed data, then compare search performance under consistent conditions.
Google's recrawl guidance says crawling can take days to weeks and does not guarantee inclusion. That is not a universal countdown from submission to final indexing. Guangsuan's SEO service connects technical, content and wider project work; the next action should follow the observed cause rather than a promised sitemap “success rate.”