tedster - 6:41 pm on Jan 10, 2010 (gmt 0)

The big bug is that site: operator is even less accurate at reporting on how many URLs are in the index than it used to be - and I'm no longer sure what number the /* hack reports on at all.

The only certain way I know of to see if a URL is really indexed is to search on the URL itself and see if Google gives a result. Otherwise, if you use site:example.com you will just get some number, often alarmingly low. And if you do individual searches on site:example.com/directory/ for all the directories, you will find URLs that were not in the simple site:example.com results - sometimes MANY such URLs.

When it comes to the primary index over any supplemental indexes or partitions, the most valuable information for me is to see what Google exports to AOL search. And with AOL hinting that they may drop their Google Search partnership [webmasterworld.com] that information source may be on life support, too.

The most valuable metrics for me come from whether a URL is getting actual search traffic from Google or not. Search results and rankings are now so personalized, localized and erratic in other ways, that only traffic itself gives a solid picture.

If a URL gets no search traffic from Google but I consider it worthy of being a site entry page, then I'll investigate whatever I can learn about its ranking and indexing as part of a debug process. Same thing if a page's traffic seems low or has changed significantly. However, the site: operator or any single ranking report, taken as primary data, can mislead an analysis rather dramatically.