Start with one representative URL and its evidence. A page loading in your browser does not establish that a search engine has fetched, rendered, selected and indexed it. “Not indexed” describes an outcome; the next action depends on the underlying state.
Separate the stages before prescribing a fix
Scroll sideways to read every column.
| Observed state | Evidence to collect | Useful next investigation |
|---|---|---|
| URL unknown / discovered, not crawled | Internal links, sitemap inclusion, inspection date, server logs if available | Discovery paths, crawl access and whether the URL is a useful intended destination |
| Fetch blocked or failed | HTTP status, robots rules, noindex, verified crawl logs | Remove unintended access errors; do not use a spoofed user-agent response as a verified crawler test |
| Browser content differs from fetched/rendered page | Initial HTML, rendered inspection and resource failures | Rendering dependency and whether buyer content/links are available |
| Crawled, currently not indexed | Dated inspection, selected canonical, actual content and similar URLs | Content selection, duplication and page purpose; a request alone does not settle inclusion |
| Alternate / duplicate canonical | Declared and Google-selected canonical, redirects and internal links | Whether consolidation is intentional or conflicting signals send value elsewhere |
| Indexed but hard to find for your query | Inspection confirms index; query/page search evidence | Relevance and ranking, rather than treating it as an indexing failure |
Build a small evidence worksheet
Record the URL, intended buyer job, first known publication date, last meaningful update and current HTTP status. Add robots/noindex findings, canonical declarations, selected canonical where available, internal-link route, sitemap entry and the content visible in initial and rendered HTML.
Record when each observation was made. A live test and a previously indexed version may differ, so one successful live fetch is not proof that indexing has already changed. Mark unavailable private inspection or verified logs as unknown.
What the E-GMP investigation supports
Strata’s E-GMP work examined migration and indexing paths alongside a dated decline: 60,489 impressions in September 2025 versus 7,687 in June 2026. Five pilot pages were implemented in July. The case explains the sequence and observed implementation/indexing evidence.
The approximately 87% decline is the problem discovered, not recovery produced by Strata. It does not establish that every missing page had the same cause. Employer buying intent also differs from jobseeker traffic, so eventual visibility alone would not settle the business outcome.
Make the correction match the finding
Fix an unintended access block at the responsible layer. For a rendering issue, work with the developer on the content and links the engine can read. For contradictory canonical signals, align the intended destination, redirects and internal references. If several URLs answer the same job, evaluate consolidation and buyer usefulness before adding more copy.
Do not make every excluded URL indexable simply to reduce an exclusion count. Filters, duplicates and utility pages may be excluded intentionally. Prioritize useful commercial destinations, confirm implementation and monitor dated inspection and search evidence afterward.
Primary guidance: URL Inspection, canonicalization and JavaScript SEO. Index eligibility does not guarantee selection or ranking.
For implementation, see technical SEO. If the problem followed a launch, use the existing redesign investigation guide and migration scope.