Reading the index coverage report without guessing at it

7 days ago
3 min read

Updated: 3 days ago

Introduction

Pages are missing from search and nobody knows why. Theories circulate, content is rewritten, addresses are resubmitted, and the actual explanation has been sitting in a report the whole time, stated in a single phrase per page.

The report is not hard to read once the statuses are translated into ordinary language. Each one means something specific, and most of them point directly at either a setting to change, a page to improve, or nothing to do at all.

1. Reading the index coverage report starts with two numbers

Begin with the shape of it.

The report divides every address it knows about into indexed and not indexed. The proportion between them, and whether it is moving, tells you more than any individual page does.

2. Expect a large not-indexed count and do not panic

The most common false alarm.

Most sites have many addresses that should not be indexed: redirects, filtered views, tag archives, duplicate variants. A large number here is normal, and the question is which of them you actually wanted included.

3. Understand crawled, currently not indexed

The status that causes the most confusion.

The page was visited and not included, which is a judgement about value rather than a technical fault. Resubmitting achieves nothing; strengthening the page, or linking to it from somewhere that matters, is the only route.

4. Understand discovered, currently not indexed

A different problem with a similar name.

The address is known and has not been visited yet, often because the site has many low-value pages or is slow to respond. This one points at site structure and server performance rather than at the individual page.

5. Treat excluded by a no-index tag as a decision to verify

Frequently correct, occasionally disastrous.

Some pages should carry this tag. The task is checking that no page you care about appears in this group, which happens after migrations and after somebody changes a template setting.

6. Read duplicate statuses as consolidation

Not a fault to fix urgently.

An address reported as a duplicate with a different canonical chosen means the search engine picked another version. Where it picked the version you intended, nothing is wrong; where it did not, the pages are too similar and need differentiating.

7. Investigate server errors immediately

The only genuinely urgent category.

Addresses returning server errors mean pages that failed when visited. Even a small number is worth checking straight away, because it can indicate an intermittent hosting problem affecting real visitors too.

8. Use the inspection tool on individual pages

The detail behind the summary.

Any single address can be inspected to show its status, its canonical, when it was last crawled and what was rendered. That is faster than reasoning from the aggregate report when you have a specific page in mind.

9. Watch the trend rather than the daily figure

Where the real signal lives.

A steadily rising indexed count on a site that publishes regularly is healthy. A gradual decline, or a sudden step change on one date, is worth tracing to whatever was deployed around that day.

Export the report and keep the file. Comparing this month's list with one from six months ago is far more informative than either alone, and the interface only retains a limited history.

Conclusion

Start with the split between indexed and not indexed and whether that ratio is moving.

Expect a large excluded count and identify only the pages you actually wanted included, treat crawled-not-indexed as a value judgement requiring a better page, treat discovered-not-indexed as a structural and performance signal, verify that no important page carries a no-index tag, read duplicate statuses as consolidation and act only where the wrong version was chosen, investigate server errors immediately, inspect individual addresses when you have a specific question, follow the trend rather than daily figures, and export the report so you can compare it later.