Ranking on google.com is different than ANY ip address or datacenter?

Am I going nuts?

[webdude]

Okay,

I periodically check some of the Google datacenter tools out there and I am totally confused right now...

One of my sites is ranking #2 for a fairly competative keyphrase when I connect to www.google.com, it has been there for the past couple of weeks. I went to a site that checks multiple datacenters and every one I checked showed us at #1. Well... that's good ;-) In fact, I checked many, many datacenters by IP and I cannot find any where I rank less then #1. So, I decided to netstat the connection when surfing to [google.com...] to see what IP was being used. Now I am totally confused. Sometimes, it shows 2 IPs sometimes 3. When I type any of these IPs and do my search, I come up #1. The IPs will vary, but every time I run the search by IP (any IP that shows up in netstat) this site comes up #1. I have tried this on several machines and I still get the same results.

Can someone please give me a little insight as to what is going on? I used to be able to netstat and get the IP that G was using, but now I am totally confused. The only time I rank #2 is when I connect to www.google.com! Whenever I connect by IP... any IP I can find, I rank #1!

[LaserEyes]

<this message ws spliced in from another location>

Google.com SERP Different Than Same Datacenter IP

I've found some behavior I cannot understand. A keyword search at www.google.com provides a different SERP than the same keyword at the IP of the datacenter being used.

I am in northern California and when I search at www.google.com for our main keyword, our website is positioned #11 in the SERP. I then ping www.google.com and get this result:

C:\>ping www.google.com
Pinging www.1.google.com [66.102.7.104] with 32 bytes of data:
Reply from 66.102.7.104

Several pings provided the same result. It appears that www.google.com searches switch to a query on www.1.google.com and are being directed to the 66.102.7.104 datacenter.

When I go directly to the 66.102.7.104 datacenter and repeat the search on the same term, our website is positioned #7, a raise of four positions and into the top ten!

http://64.233.179.104/search?hl=en&q=keyword will position us at #7
http://www.google.com/search?hl=en&q=keyword will position us at #11

Then I dropped the www from the ping.

C:\>ping google.com
Pinging google.com [64.233.187.99] with 32 bytes of data:
Reply from 64.233.187.99

The datacenter IP has changed to a different C-Block, however if I go directly to 64.233.187.99 and query the same keyword, our page is in position #7.

http://64.233.187.99/search?hl=en&q=keyword will position us at #7
http://www.google.com/search?hl=en&q=keyword will position us at #11

I commonly use <an online> Google datacenter watch tool to follow our rankings for keywords. It is showing position #7 at all of the datacenters they include. If I go directly to the IP address of any of the datacenters they include and duplicate the search, I get the same result. We are #7.

I had a friend in Los Angeles do a www.google.com search on the same term. We show up as #11.

I shut down my DLS and went dial-up to test if there was a difference. None.

I used a proxy server in a foreign country and repeated this process. Results are the same. If you use www.Google.com we are at #11, if you use the IP of the datacenter indicated being used by pinging, we are in position #7.

Is this known behavior? To my knowledge I've never seen this happen before.

Is there something I've done to cause this?

Is there something I can do to stop us being dropped four positions for www.Google.com searches?

<Sorry, no specific keywords.
See Forum Charter [webmasterworld.com]>

[edited by: tedster at 2:14 am (utc) on Nov. 9, 2006]

[webdude]

Yep LaserEyes,

I am kind of seeing the same type of behavior. I think it may have to do with a redirect. What is wacky is that no matter what IP I use, I pop up #1. When I use www.google, I pop up number 2. I am hoping the #1 eventually sticks. All indications look good.

[300m]

Whats really interesting is running the netstat on a clicked cached link. I get 2 additional ip addresses. One being the cached IP and another being a different IP address other than the default Google IP.

[theBear]

Don't you just love massive dynamic musical server chairs ? ;-)

[randle]

Google SEO tools that have gone away;

Back link checking
Similar page function
Tool Bar Page Rank
Understanding what your position is by looking at data centers

What is described above we have definitely experienced. If you go to your favorite DC checking tool, you can never find there what you see when you just pull up google.com. Sounds crazy, but I don’t think you can really tell where you rank anymore and that’s a little frustrating (perhaps that’s the idea?). Sure, there was always some variation across data centers but generally speaking you could evaluate your position, and get some sense if you were moving up or down.

If you make a small optimization adjustment it’s very difficult to tell what effect it had anymore. Sure, you can see if you have moved from #60 to #15, but try and pin down moving from # 8 to # 4; very challenging these days.

[300m]

I guess this is what Matt was talking about back in april at Pubcon. I left that session with the impression that Big Daddy will leave dataceter watching a thing of the past. With regards to SEO tools gone, there is mention of a keyword tool that is throwing back 502 errors when an API hits the IP. Maybe something within the infastructure has changed recently.

[g1smd]

What result do you get if you use the equivalent GFE datacentre name instead of the raw IP address?

[webdude]

g1smd,

Just a quick spot check on these show the exact same results. #1 for the IP, #2 for the following...

[gfe-ar.google.com...]
[gfe-fg.google.com...]
[gfe-po.google.com...]

[webdude]

As of today, I am still seeing the same result here. Anyone?

[steveb]

Not sure what you are saying, but datacenters gfe and numeric haven't been the same as google.com for many weeks now.

[webdude]

It isn't just gfe datacenters, it's any datacenter I can find. Every datacenter I check shows this site #1. Everytime I check google.com and show the IP and try that, same result - #1. Every time I ping google.com and try that IP - same result - #1. I have gone through long lists of these datacenters from previous posts. Any type check I do shows this site #1.

Except...

When I run the search on google.com, the site is #2.

mmmmmm... Where is that coming from?

In fact, I cannot find a datacenter where I do not rank #1. Only when I search google.com.

Is this making sense?

[Gissit]

It seems possible to me that the datacentres supply a set of results to google.com that are then crunched to form the SERPS. Where does the site that ranks #1 on google.com rank in the datacentres?

It would not seem so odd that the datacentres do not house the entire algo and it seems quite feasable that there is a further set of data elsewhere (not available to you or I) that is factored into the results of the datacentres to build the served SERPS.

Hypothetical, yes.
Unlikely, maybe.
Impossible, certainly not.

[edited by: Gissit at 6:15 pm (utc) on Nov. 16, 2006]

[steveb]

"Is this making sense? "

Well yeah, as I said, the datacenters haven't conformed with Google.com in several weeks. There basically is no reason to look at the datacenters anymore in terms of ranking itself.

[manca]

It's very weird ranking for my site too.
When I go to www.google.com and search for my kwords, I can't find my site anywhere? Even not on 10th page....The site has used to be number 1 for more than 2 months now... The weird thing in the whole story is that when some other people (my friends) from different places in the world check google they can still see my site at the 1st position. Another thing that scares me off is when I use GFE for searching my site still shows on the 1st position.

What is happening with big G these days?

[AlexK]

webdude:

Any (datacenter) check I do shows this site #1
.
Except...
.
When I run the search on google.com, the site is #2.

Are you suggesting that G is stuffing the google.com #1 site in above yours?

(...now ducks, and runs...)

[webdude]

No... I am not suggesting that. I just don't get it. Obviously the data being pulled for google.com is something that we can no longer access. Searching datacenters, any datacenter, seems to be pointless. Granted, we are just talking about a ranking difference between #1 and #2, but if G made the datacenter searches pointless, we may as well chuck all the "Datacenter Watch" posts out the window. Not to mention the sites out there built for multiple queries of these datacenters.

Might just be an anomaly for this particular site, but I have been trying for the #1 spot for a couple of months now and was just wondering what was up.

mmmmmmmm

[caryl]

I don't know if this will make any difference but...

I have noticed that when I go to google via an IP address, my "preferrences" are NOT saved.

If I have my preferrences set to display more than 10 results per page - the results ARE different than when I just search for the default 10 results per page.

[BillyS]

>>I periodically check some of the Google datacenter tools out there and I am totally confused right now...

I think that Matt Cutts even mentioned in his blog that datacenter watching can be a tricky thing. With all of the DCs needed to support Google it's not so black and white anymore.

For example, I know of at least one tool that used to be a very good indicator of ranking that is now almost completely useless. I used that tool to track rankings and now the results I find there I cannot find anywhere else.

[300m]

I agree about the datacenter thing. When i ran that netstat on the day you posted this thread and I noticed there were 3 different dc ip addresses, I realized it's changing in the sense we cant say "check this out". I too use a reliable tool that polls a data center with an api and I dont trust it as much as I did 2 weeks ago.

[webdude]

If I have my preferrences set to display more than 10 results per page - the results ARE different than when I just search for the default 10 results per page.

Interesting... I am seeing the same thing. In fact, when I switch my preferences to display 100 results, I get indents. When I set it to display 10 results (the default) I am getting no indents.

mmmmmmmmm

[g1smd]

It may be that the google.com results take your location and/or language into account in some way that GFE or IP access does not ... or it may be some other factor.

[LaserEyes]

I have not been able to duplicate the change in positioning in any manner other than using the www.google.com URL. All IP addresses show us as 7, whereas www.google.com show us at 11. The www.google.com does have the indented subpage.

Our Google Webmaster Tools search query stats show us with an average high position fo this keyword as 8 and this keyword is our number one clickthough.

If you think it is frustrating being bounced from #1 to #2, consider that we are bounced off the first page. Aaaarrgh!

[tedster]

when I switch my preferences to display 100 results, I get indents.

That's because Google indents are generated by clustering two results that would normally appear at different positions on the page. So if you're set to 10 results per page, and a given domain has results at #3 and #9 in the "raw" results, Google will cluster them at #3 and #4 in the final display.

Now set your preferences to 100 results per page, and if some domain would have results at #3 and #87 in the raw results, Google will cluster them at #3 and #4.

The nice thing is that if your domain is at #1 and #11 (set to 10 per pge) then all you need to do is nudge that second page result up ONE position, and you see it displayed as #2, not as #10

[webdude]

tedster,

Mmmmmmm... I never tried that. I might want to pay attention to those second pages. Thanks for the tip.

[activeco]

It may be that the google.com results take your location and/or language into account in some way that GFE or IP access does not ... or it may be some other factor.

In my earlier tests I found different results depending on many factors:

-my IP
-my Google search preferences
-Cookies
-Toolbar Server
-Language preference in my browser (very influental)
-... can't recall the rest, but sometimes I had the feeling that my recent search behaviour played a part of different results too.

[tedster]

recent search behaviour played a part of different results

If you're "logged in", most definitely