When is an indexing repair complete? Our 18-URL follow-up
An indexing repair is complete when the exact URL reaches its intended state in a later inspection. API acceptance is a submission result. Our 18-URL sitemap had 9 non-indexed URLs on September 25, 7 on September 28, and 6 on October 8. We cannot attribute the change to any submission or link edit.
Which URLs changed in the same sitemap?
We compared all 18 URLs, nine English and nine Chinese, using Google URL Inspection. The table follows the nine URLs absent from the index at the first inspection. Dates and counts are our own measurements in Asia/Shanghai time. “Indexed” means the inspection response said “Submitted and indexed”; it does not measure search traffic.
| URL path | Sep 25 | Oct 8 |
|---|---|---|
| /services | Discovered, not indexed | Discovered, not indexed |
| /zh/services | Unknown to Google | Discovered, not indexed |
| /zh/about | Discovered, not indexed | Indexed |
| /zh/case-studies/pet-tech-dtc | Discovered, not indexed | Indexed |
| /case-studies/welding-dtc-delivery | Discovered, not indexed | Discovered, not indexed |
| /zh/case-studies/welding-dtc-delivery | Discovered, not indexed | Unknown to Google |
| /zh/privacy-policy | Crawled, not indexed | Indexed |
| /terms-of-service | Discovered, not indexed | Discovered, not indexed |
| /zh/terms-of-service | Discovered, not indexed | Discovered, not indexed |
The full-sitemap numerator fell by three, with the denominator fixed at 18. The English and Chinese service pages, both welding case pages, and both terms pages still need investigation. A changed coverage label is an observation at inspection time; it does not identify the cause.
What did the live repair check rule out?
On October 8, each of those six URLs returned HTTP 200, resolved to itself, exposed a self canonical in raw HTML, and had no robots noindex meta tag. The six are still in the sitemap. We fetched raw HTML from the 12 URLs marked indexed in the same inspection and counted distinct source pages linking to each sitemap URL, excluding self-links:
| Target group | Targets | Indexed source pages linking to each |
|---|---|---|
| Still not indexed | 6 | 5 each |
| Indexed | 12 | 4 or 5 each |
In this small site, five inbound source pages did not distinguish the six unresolved URLs from indexed URLs. The count includes shared navigation or footer links, so it does not measure editorial prominence or explain Google's selection. These checks make an obvious redirect, canonical, or noindex mistake less likely; they do not prove that Google considers the content worth indexing.
Our static sitemap entries have no lastmod. We will add one only after a traceable page change; stamping today’s date onto unchanged pages would misstate freshness. The next investigation is page-specific content and crawl evidence, particularly the service and welding pages. The terms pages may offer limited search value, but we have not chosen an intentional noindex policy for them.
When can we close the finding?
Our October 8 submission record separates three payloads:
| Channel | Payload | Immediate result |
|---|---|---|
| Google sitemap API | Whole 18-URL sitemap | Submitted |
| IndexNow | Six unresolved live URLs | HTTP 200 accepted |
| Bing SubmitUrlBatch | Same six URLs | Six accepted, zero deferred |
These are delivery records, not proof of crawling or indexing. Bing crawl signals are partial and cannot be counted as indexed membership. GSC showed 0 clicks and 17 impressions in the settled September 29–October 5 window; GA4 organic landing pages were empty. Neither measure diagnoses why a specific URL remains unindexed.
Our acceptance check is: re-inspect the exact URL, compare its coverage state against the same 18-URL denominator, and preserve the dated response. We will close a page only after it is indexed or intentionally removed from the index with a documented reason. The cost is leaving six findings open while evidence accumulates.
Sources and open loops
The first-party artifacts are the nine-row transition table from September 25 and October 8 full-sitemap URL Inspections, the October 8 raw-HTML link census, and the October 8 channel-level submission record. The detailed response files stay in our private operations archive; the method and aggregate results are public here. The next check will ask whether Google crawled the six URLs, whether content changed, and whether the index state changed afterward. We have no measured causal answer yet.
Next step
See how we scope this work as an engagement, or request a diagnosis of your own site.