Forum Moderators: Robert Charlton & goodroi
In the past two upheavals by Google (8th March 2006 & June 27th 2006)it has been apparent that the changes have been roughly accross the board, with the obvious exception of the 64.233.189.104 DC.
The DC seems to use the same cache dates as the other DC's and has the same amount of supplemental pages also, and yet somehow it has ignored any penalties that have been applied to the other DC's and gives treats my site as though nothing were wrong.
Any idea what the purpose of this DC could be?
All the Best
Col :-)
[edited by: tedster at 7:46 pm (utc) on July 21, 2006]
1. Site loses indexed pages
2. Pages come back
3. Changes appear like titles, PR, backlinks
4. Go back to 1 and repeat.
It's not every site at the same time which is interesting. Each one is at a different part of this cycle at any given time.
72.14.207.**
I can see that I have lost ranking for some few competitive key phrases, and gained other few. Guess new filters or a new wave of "Matt Cutts - Data Refreshes" underway ;-)
From now on maybe we should start give names or numbers for those refreshes things. Something like:
Data Refresh - Emmy, for example :-)
I'm seeing some sort of a data refresh or out-and-out glitch on
gfe-ar
gfe-nz
gfe-po
gfe-ro
for a very competitive phrase that we've been #1 for quite some time, the index page dropped out of the top 1000 -- granted, it is still #1 for the plural version of this word, and maintains dominance on the allins for that initial phrase.
The only thing I could note is that for this phrase on this dataset, the number of results is a few million less than the other 37 dc blocks. Hopefully it is restored soon, as it is showing the frightening results every few minutes.
Cygnus
So long as I copied my list down correctly, that should be gfe-eh.google.com, which I'm not seeing anything weird on for the phrases I tend to follow.
I noticed something a few months back, but never spent a lot of time digging into it...when the refreshes come pouring out, they appear to be on a phrase-by-phrase basis, not site-by-site. One dc can show a site ranking very well for one phrase and then it incorporates a new dataset for a similar phrase (with the same consistent ranking history) and knocks it from the top 1000. They never seem to to fully propogate though, sometimes folding into existing dc flavors, or sometimes disappearing altogether.
There's not much I can really do but wait; 2 other pages for this site now rank for the phrase around 50 and 150, so once the index page is recrawled and recalculated, it should be back to the previous #1 (or so I'll keep telling myself).
Cygnus
[edited by: JoeSinkwitz at 9:55 pm (utc) on July 21, 2006]
After about 24-48 hours, my site was fine. In the meantime, I was sweating bullets trying to figure out what was going on. Adding &filter=0 brought back the appropriate #1 result, adding "" brought back the appropriate result, and so as I saw things start to improve for me, I made a post. I suppose I should do it again...
"Steveb's 75% database theory seems to be the most plausible at this point, and it really is on a query-by-query basis as the data is torn and rebuilt. I've now watched several big terms cycle through the DCs for established sites disappearing and then popping back in.
For those with a vested interest in the outcome, I don't think watching the DCs this weekend is going to have a positive effect on your sanity. For phrases that I saw disappear and then come back, the overall quality seems a bit higher, and assuming that Google's goal is more relevancy (tinfoil hats aside), I don't think you'll have too much to worry about when all is said and done.
With so many operators gone wild and various craziness when adding the &filter=0 or using ""s, we know it isn't done cooking yet. The best I can offer is that the magically delicious disappearing, reappearing results HAVE been coming back. Disappear from the DC watch this weekend; I'm sure next week will be equally exciting for us."
Of course, it is off by a day, so I'll probably end up watching the DCs closely, but for those of you that just watched a result disappear completely, try the two operator tricks, breathe deeply, and know that you're not the only one seeing this.
Cygnus
How would that account for still ranking in problematic datacenters for equally, if not more competitive queries?
I will agree that their concept of aging and authority isn't all it's cracked up to be, but I'm not sure that is it -- although I will admit that when this happened last time in the DC watch thread around June 14th, those that exhibited the same problem had sites that were at least a few years old (probably more a byproduct of only older, established sites making that kind of headway in the first place).
Keep the ideas coming; I'm curious why this data drop and rebuild is happening every 5 weeks, and even more curious why the &filter=0 or use of "" bypasses it.
Cygnus
[edited by: JoeSinkwitz at 11:02 pm (utc) on July 21, 2006]
64.233.189 - Anyone got any comments on this DC? To me it looks good for our sector at least.
This DC remains pretty much as it was before 27th June. It's the one DC (that I know of) that has not had the 27th's bug-ridden code applied.
I'm guessing that Google are holding this DC back as a control for whatever hair brained disaster they introduced on the 27th June. We can only hope that they've realised their mistake and will now use it as a control for the bug-fix.
Ran a subset of 5 terms 1-10 rankings on one domain (we manage 20+ campaigns) and everything was as usual for me.
Granted I haven't ran EVERY site we monitor through it, but a stable one with top 10 rankings on comp of 400 Million plus, runs same as ever.
Ran a subset of 5 terms 1-10 rankings on one domain (we manage 20+ campaigns) and everything was as usual for me.
That's exactly what I would expect you to see. The 27th bug didn't make any discernable difference to the rankings - other than surgically penalising certain sites and dropping them 40,50, or 60 places. All of the other sites' positions remain exactly the same relative to each other.
I am seeing differences between my keyword and "my keyword".
"Watching G lately is like watching a crazed fly buzz around the room."
Well said!
And its therefore no Google employee wish to comment on current situation or at least post a weather report as they use to do in the past, IMO.
Or at least to post something like:
Damn we screwed up big-time :-)
Interesting after all of this time that there is still some good data running around out there.
64.233.189.104 is the only stable one that I have seen since 27th June also. It seems to use the same new data source as the others, but hasn't applied new criteria and so I have missing pages and supplementals on it, but good positions for the pages that are left.
I am seeing about a 1/3 return on the other DC's and supplementals are just down to about 25 pages now, but still when I site: search my 1st page is a mid range background one, instead of my home page.
All the Best, It's looking like Barbeque weather here ...
Col :-)