The Google Search Console indexing report is a trend tracker, not a repair queue. That framing comes straight from Google. In episode 112 of Search Off the Record, Martin Splitt and John Mueller spend an hour on one point: the report exists to confirm that your intentional changes are processing and to surface systemic patterns.
It is not a list of defects. Most sites that read it as one end up “fixing” things that were never broken.
Key takeaways
- Read trend lines, not row counts. A steady pile of not-indexed URLs is normal on every site.
- A 100% index rate is not a goal. Google says its own developer docs carry a large chunk of non-indexed pages by choice.
- Set up a domain property. It folds www, non-www, http, and https into one clean view.
- Trust Search Console over the site: operator, which can show stale data for years.
What is the page indexing report actually for?
Google’s Search Console team built the page indexing report to answer two questions. Did the change you made on purpose get picked up? And is anything moving in a direction you did not intend?
Deleted a few hundred old pages? Watch the expected 404 line rise, then settle. Migrated domains? Watch the new property fill as the old one drains.
The report earns its keep during change. On a quiet site it should mostly be left alone.
Why is “not indexed” not an error list?
Because the not-indexed bucket mixes correct behavior with real problems, on purpose. A page you deleted returns a 404, and that 404 shows up in the report as a not-indexed page. That row is your server working.
The same goes for pages you set to noindex, URLs that redirect, and canonical variants Google folded into another page. All of it lands in the same bucket as genuine issues.
Here is how the common states break down, and how much attention each one deserves.
| Report state | What it usually means | Panic level |
|---|---|---|
| Excluded by noindex | You told Google to skip it | None, if you set it |
| Not found (404) | Deleted content answering correctly | None, if deletion was intentional |
| Page with redirect | The URL forwards somewhere else | None |
| Discovered, currently not indexed | Google knows the URL, has not fetched it yet | Low. Watch the trend |
| Crawled, currently not indexed | Google fetched it and chose not to keep it | Quality question, not a bug |
| Server error (5xx) | Your host failed on some requests | Low for blips. High if it grows |
Do you need a 100% index rate?
No, and no magic ratio exists. Mueller is direct about this in the episode: there is no healthy percentage of indexed to non-indexed pages, and chasing one wastes time.
Google’s own developer documentation site carries a big block of non-indexed pages because the team behind it chose that on purpose. Old API versions, redirected paths, and utility pages sit out of the index by design.
The number that matters is different. Are the pages you care about indexed, and do changes you make show up over time?
The wrong way
A checklist to clear
- 1,204 pages not indexed
- 310 expected 404s
- 95 noindex tags you added
- 12 redirects working fine
Endless. Most rows are correct.
The right way
A trend to watch
Step change: look here
One question: what moved, and when?
What does a healthy report look like after a migration?
Slow, then fine. Search Console data lags by a couple of days, and reprocessing a full site move takes weeks, not hours.
The “validate fixed” button helps here, with one caveat. It is a tracking tool that asks Google to recheck a batch and report progress. It is not a lever that forces indexing.
Press it after you change something real. Then let it run its course, which can take weeks on a big site.
Why is a domain property better than a URL prefix?
A domain property covers every variant of your site at once: http and https, www and bare domain, and every subdomain.
That kills a whole class of ghost problems. When Google shifts its canonical choice between your www and non-www versions, a URL-prefix property shows traffic vanishing from one and appearing in the other. A domain property shows the truth, which is that nothing happened.
Can you trust the site: search operator?
Not for diagnostics. The site: query is an artificial lookup that skips most of Google’s normal serving logic, and the episode gives examples of it showing traces of old domain moves and hreflang swaps for years after they were cleaned up.
If Search Console and a site: query disagree, Search Console wins. It reads from the same systems that build real results.
The same discipline pays off beyond Google. The crawl and index habits you build here are the ones AI answer engines reward when they pick citations.
FAQ
Should I press validate fixed on every issue?
Press it when you changed something and want Google to recheck. It tracks reprocessing and emails you the outcome. It does nothing for rows that were already correct behavior.
How many of my pages should be indexed?
There is no target number. Google’s own team says even their documentation keeps a large share of pages out of the index on purpose. Judge the pages you want ranking, not the ratio.
Do temporary server errors hurt my rankings?
Single failures do not. Computers fail, DNS requests drop, and Googlebot just retries later. A growing 5xx trend over days is the pattern worth chasing with your host.
Where do I start if my report looks scary?
List the ten pages that earn you money, and check those first with URL inspection. If they are indexed and clean, the scary numbers are almost always correct behavior in bulk. If you want a second set of eyes, we read these reports for a living.
