Google crawling old index - the reason for a traffic drop?

[londrum]

i had a bit of a traffic drop recently because of this panda thing, and my number of indexed pages has halved. But after digging around in my stats and WMT, i noticed that google has been crawling a load of non-existant pages.

im talking about pages which existed on my site at least 12 months ago, which i deleted, and have never shown up as errors in WMT before. now all of a sudden they are back and i've got bazillions of 404 errors.

WMT shows you were it got the URL from, and some of them are external sites. If i follow these URLs then more often than not the link does not appear, so im guessing google has crawled on old version of that page too, and the link has since disappeared.

so i'm thinking that might be part of the reason for our traffic drop.

google has crawled a very old index for panda, and is doing the same for all the other sites, so all our backlinks will be screwy, and all our internal links too, and all our new pages have yet to be indexed.

...even if your site has been thoroughly indexed, you might want to site it out before making changes, so google can factor in all the other sites too.

[netmeg]

There is definitely something a bit odd going on with old URLS in GWT - I completely changed the entire URL structure on one of my sites about four months ago, and put 301's in, and Google had it nicely sorted out, and most of the old URL 404s dropped out of GWT and everything was everything till a couple weeks ago, when suddenly out of the blue around 400 old URLs popped back up as 404's. No idea from whence they came - they *were* gone.

[indyank]

netmeg, I am spending a lot of time on GWT these days because of such issues. I even found an answer to a question that several were asking here for long.

Why does google ask us to confirm both www and non-www versions for only some of the site and not all in GWT?

It looks like Google does this only when you try to set the preferred domain option to one of the versions.For sites that do not set the preferred domain option, they aren't asking to verify both the versions as they treat them as separate sites!

[netmeg]

Right, I knew that. I have around 500 sites in my GWT because of that.

(They treat them that way because technically they *could* be separate sites, with different people in control of them. www is just another subdomain, after all)