How indexing works
Every Capable capability writes what it knows about a page into Confluence's search index. That write happens in the background, on a schedule, not the moment you save a page.
Understanding that explains almost every surprising search result.
#What indexing produces
One stored summary per page, written to by every installed capability. That summary is what backs the Capable columns, filters and sorts.
#When it runs
On a background schedule, continuously.
On demand, when an administrator runs it from global or space settings.
A large site takes hours to days to converge from a standing start.
An unindexed page does not match, and does not appear as an empty row. It is simply absent. That is why a filter on a fresh install can look like it is broken when it is only early.
#Forcing it
Administrators can run indexing from the global settings Indexing tab, and space administrators from the space settings equivalent, which also shows coverage for that space.
#A few things worth knowing
Index coverage is the number to check before concluding a search is wrong.
Re-indexing a space is the usual fix when one space looks wrong and others do not.
Indexing is per page, so a page nobody has touched since install may still be missing.
#Related
Indexed, then searchable. In that order.
