Strange January Keywords - data from Google Analytics

[Gorgwatcher]

After this may day algo, we have been busy in digging our GA data and comparing the stats.

During this activity we have come across a very strange behavior(or bug) in GA.

In jan, our keywords data shows some long tail keywords from Google that are totally illogical but gave us some good traffic means above 60 per month.

But in preceding months like feb, march, april, may, the very same keywords give 0 traffic.

And the strange thing is these keywords are still on 1 st page (some are on 1st position even)

These are keywords with 4 5 words (some are 7 worded) And
we are still on 1st page in Google on these terms but with 0 traffic

Has anyone else notices this thing? any logical reason?
Also they are not seasonal keywords.

we are considering another factor of traffic loss,which happened to so many websites, that may be Google has enhanced its Analytics Software.

Previously we have 60 plus visitors from the keywords(5 worded, 6 worded) that only one person can think of. but now they seems to appear less.

anyone else saw this thing in GA?

[tedster]

Once in a while I've seen a batch of very strange keywords in a GA report - and not just in January. I don't think those were a GA issue. In cases where I can access the raw server logs, I do see that keyword traffic. So it's more likely to be a very strange ranking problem in Google itself. The SERPs can get mighty strange at times!

[peterdaly]

I sometimes wonder if it's automated tools trying to detect cloaking.

MSN did this a couple of years back, sending "fake" traffic from unrelated keywords to (I think) try to detect cloaking.

[zehrila]

To what i understand, i guess it could be related to some news or some local event, which triggered traffic for short term?

[leadegroot]

I've seen some very strange keyword combos (that I don't believe a real person would use) in GA - and they *don't* appear in my raw logs.
I don't know what Google is doing, but I don't put a lot of faith in Google Analytics :(

[tedster]

Thanks, leadegroot, now we've got two reports.

It sounds like something might be going wrong in Google's back end, doesn't it? The idea that GA would report any visit that isn't in the raw server logs is unsettling.

[leadegroot]

This has been going on for a long time now - I don't have any recent reports as I haven't drilled into that area recently, but some 6 months ago I had 60 instances of a keyword phrase that was... strange. Not impossible, just people don't talk that way. This was across a month and not once did it appear in my logs. I checked carefully because I was so dumbfounded to see so many instances.

The high level of GA matches my other reporting efforts, but the detail is sometimes just off.

[tedster]

I wonder if any of it is Google Search traffic that does NOT send the referrer, hence no keywords for the server logs. For example, the new Google encrypted search [webmasterworld.com]. Or possibly some new ajax SERP experiment gone wrong, as one did a year ago.

[leadegroot]

That... is possible, The specific term that struck me as so wrong that I have remembered after all these months would be most likely to be real if it was coming from overseas, where the terminology is different.
This site is .au based, so if that group was from the US, where things are more likely to be tested anyway... that would make sense...

I would not like to see a situation where a webmaster could only get real information on what traffic was happening by being locked into Google tools.

Lea

[tedster]

I would not like to see a situation where a webmaster could only get real information on what traffic was happening by being locked into Google tools.

That's been on everyone's mind since the Ajax fiasco a year ago. It would be sooooo heavy handed if something like that went down. To get webmasters to accept it would take more spin than even BP needs.

I don't want to borrow any trouble here, so I'm assuming it will not come to pass.

[Gorgwatcher]

@zehrila
I have already mentioned these are not some seosonal or event based keywords

@leadegroot
Yup they indeed are simply strange.
for instance I received above 60 hits from a 7 worded keyword :S
Which means either a same person is coming back to us by searching that 7 worded query in google for above 60 times in a month (seems totally illogical) and then that specific person dies as the hits drop to 0 after jan :P

And teh issues is they still are on first page of Google but with 0 hits

So for us, Jan is the troubled month

[tedster]

above 60 hits from a 7 worded keyword

That really sounds like automated traffic of some kind, doesn't it.

[Gorgwatcher]

well yes indeed it seems. BUT we have keywords in thousands with this odd pattern in Jan. Lets say from a sample of 500 Keywords, 266 keywords are of this nature. Which gave us traffic ion Jan but after jan they simply drop to dead 0. And they all are still on first page. These keywords are like

"item country best quality wholesale price"

[peterdaly]

Is all the traffic from client IP addresses on a similar subnet?

If so, find out who owns the subnet, and see if that indicates anything. (MSN is one who has been known, in my personal experience, to send bogus requests that look like live users.)