I've just been reading some good articles on Google's personalization and it looks like the old methods of SEO and site rankings is pretty much out the window. They claim this will bring us better targeted traffic...so far I'm not seeing that effect.

Reno: It's possible that some sites are feeling this more prominently than others

Exactly - this is not a widespread phenomenon and it never was. It's something that is only rarely seen but when it is, the numbers a dramatic.

All the well-intended but academic questions and ideas are not very illuminating, because we're not getting input from many members who are working with the thing itself, in real time.

The test is to methodically check position for a small handful of keyword phrases at regular intervals

I think some proxy checking might also be in order to see geographic variations. Or, even better than a proxy (whose IP is usually well known to Google) would be if you have access to a VPN and can use that to check.

I dont know exactly if this is a throlling issue but my blog is experiencing a +8/+10% traffic increase every day since last monday. If I look at my Google Analytics graph it looks like the Everest Mountain (lol) and i dont know if being happy or worried.
Weird enough my sales (and clicks) are fairly the same ..is it possible that all these new users are all so smart not to click or buy my items?

This increase started a few days after I enhanced my blog by improving my theme and my blog speed. I used sprites in my theme and i checked my blog using firebug. i fixed a few things and i made my blog faster (under 5 secs to download it). dont know if this thing is related to what i am experiencing though...

Hello frank72, and welcome to the forums.

The only way that what you see could be related to traffic throttling is if your traffic was unnaturally flat for an extended period of time and this is the way that you are seeing the throttle being removed. Was your Google search traffic flat?

Hello and thanks for your reply. If, by "Flat Google search traffic" you mean that my Google traffic was "as expected" over a determined period of time my reply is Yes. Before the spike I am experiencing, my Google traffic was as regular as ever.
Starting from monday, all of a sudden, my google traffic has started increasing regularly.
Overever, looking at the "unnaturally" term you just used I would say, No, my Google traffic was not unnaturaly flat. My blog has been growing regulary over the last three years. It has been doubling year after year.

Here is some data that Drall provided earlier in this thread:

- PR 8 site.
- 1.1 million searchs per month.
- 120,000 pages handbuilt over 12 years.
- 36,000ish searchs per day

Looking at the data, it appears that a large majority of the pages on the site don't receive any visitors at all from Google on any given day. Also, does the site contain 120,000 pages of unique relevant content on a single subject? Do all the pages immediately drop out of the SERPs when the daily limit is reached, then reappear early the following day?

Hi Aristotle,

Most of our traffic is what Tedster calls "fat belly" 2-4 word phrases and longtail 4-10 word phrases. About half of our pages are in the primary index 60,000ish range. All of the content is unique handbuilt relevant content to one subject but covering hundreds of subtopics of that category.

Here is what happens.
-----
(X refers to pages in my site with stable daily refers, Y refers to a page that spikes)
-----

Everything is humming along normally on Monday, all X's (thousands)are getting exactly what they usually gets traffic wise so far into that day.

All of the sudden we have a Y that ranks for a major traffic phrase due to social networking, viral, IBL...

Lets say X's usually shows at position 2-10 for serps.

After Y has spiked all X's get depressed to positions 8-25 for the remainder of that day to balance out the difference of what Y spiked for.

If Y continues to spike and stays at that new level of refers then all X's stay at those new depressed rankings forever.

That's basically what is happening.

That's it, thank you! It sounds like the smoking gun to me. We know from a patent filing that Google has mentioned the idea of tracking domain impressions (not just individual URL impressions). Just extend that to domain "clicks", but add in some ranking effects as a result.

Would be great if others can look for the same thing - I think the fat belly search terms may be an important part of this picture, rather than head terms or long-tail. And now the further question comes up - why would some domains be given this treatment?

Thanks again drall for the detailed account you've provided in this thread ~ you've made the point convincingly (to me at least) ~ now we need to figure out what to do about it!

....................

why would some domains be given this treatment?

I understand tedster's request for hard data, but for the question above, it's not the worst idea to theorize a bit, to see the possibilities.

Could it kick in because some queries bring back a lot of perfectly acceptable results, so this "throttling" is G's way of giving some of those equally-valid sites more visibility (that they otherwise would not get)?

Is it more prevalent with the hottest keyword phrases? or will relatively obscure searches also trigger the same thing with certain domains?

Does it happen at approximately the same frequency to info sites & affiliate sites & commercial sites & social networking (such as blogs & forums), etc (in other words, "across the boards")? or is it more likely to happen, for example, to a site with a strong affiliate aspect and somewhat less original/authoritative content?

Many questions, few answers at this point...

...................

Tedster, I've seen this too on one of the most trafficked sites I'm involved with.

It's been going on for years and traffic is a flat line. Pretty high volume for our neck of the woods (3-4K a day).
However NOTHING will move this site. Tons of new links coming in every month due to a heavy load of new content being developed daily - no dice, has been sitting firmly at the same level for ages.

I don't believe it's related to page/server speed as we moved recently from a slow host to a new lightning fast server.

It's a Wordpress site, it has Adsense, it has GA and a GWC account installed.


Personally I think Adsense and Analytics might have something to do with this, but of course I'm not sure.

I still say there is no throttling mechanism per se, only the temporary appearance of such. I suspect the new personalization system is still kicking in and digesting a ton of data. The ups & downs are typical of Google every time they fire up a new (major) algo change.

The concept that Google would try to share traffic from bigger sites to lesser sites is hard to accept. Why would they do that? Only wishful thinking here me thinks.

I have noticed that my Wordpress blog is getting more traffic, but that is just supplemental info to my main site. Google seems to like CMS sites because of their clean content / css separation. I figure people will all start using CMS, then Google will decide that using a CMS is black hat...who knows. Any way you cut it (using Occams Razor, lol) the simplest explanation is usually the correct one and the current issue simply seems to be a side effect of the new personalization system, or not.... ;^D

BTW: Traffic and sales are a bit better today.

well, I too experience this...I always used to wonder why it remains so flattish and going by the discussions here and the google patent that Tedster refers too, traffic throttling based on domain seems to be definitely on...

But what is google achieving by throttling the traffic to a site? It is strange...

It is like an UCL being set for a domain, but i haven't been tracking the traffic fluctuations during the day...may be, that will make this more obvious...

I have another example as well...This particular site lost a key traffic pullng page from the google search engine for some reason...but to my surprise, the traffic remained steady...a few days later this page reappeared again and the traffic still remained the same...I definitely see a domain level throttle being applied in this case...

indy - which patent did Tedster refer to that describes throttling? Can you share a link to that info?

I don't have the kind of "smoking gun" evidence that Drall has, but I've seen things that make me suspect some kind of throttling. The clearest evidence I've seen is that deliberately reducing my ranking for some terms seems to increase the rankings and traffic I get for other terms.

As for why Google would do it - to create artificial stability. Too much traffic fluctuation would create economic chaos as business revenue would similarly fluctuate, leading to (even more) backlash against Google from the business community. It would make sense for Google to implement measures to increase stability of traffic, especially as they get more into "real time" search which would normally cause increased instability.

That patent didn't describe throttling, it described measuring impressions by domain as part of another process - and it mentioned it as a side point, no as the key focus for the patent. I didn't have the time to dig it up right now, but it goes back at least 3-4 years, I'd say.

no throttling mechanism per se, only the temporary appearance of such

By "temporary" you'd have to mean months to account for some of the observations - even the same Google search traffic levels on weekends and weekdays. Or no traffic at all after 1 pm, day after day - even though there are 100s of visits per hour until 1 pm.

And what about drall's recent post above, where a traffic spike for one keyword is balanced out by lost ranking positions on other words until the spike goes away?

I don't say traffic throttling is common, but I do say it exists in some cases. I've seen too much to think the idea is all tinfoil hat - that's why I resurrected the topic. And drall is the first person I know of who has pinned down the "how" of it.

As in the past threads, it is easy to doubt that traffic throttling exists. I've been back and forth on it myself, because it IS a very strange idea - and after all, who would want it to be true?

But the French Academy of Science once said that hot stones don't fall out of the sky, too. Just because almost no one gets to see a meteorite fall doesn't mean they don't exist.

By the way my blog is on Wordpress too. Today my traffic is a little diminished but it is still pretty high if compared to my normal trend...however by the end of the week my traffic tends to diminish.
All this is pretty weird but the pattern is something i have already seen. During my 3 years such fluctuations have already occurred but on a smaller scale...other times the traffic went back and it settled on a slightly higher volume... Looks like Google is testing sites on a kind of trial and error fashion...

If there is stability in organic serps, it means there is only one way to increase traffic through google. Adwords.

Tedster said:
And now the further question comes up - why would some domains be given this treatment?

Actually, it's possible that the Google algorithm sets an upper limit on daily referrals for all domains, with each limit based on the site's size, authority, and other characteristics. Most site owners might never realize that such a limit exists, because their Google traffic never gets high enough to bump up against it.

That is one possibility - it seems outrageous, but so does this whole throttling idea in the first place, doesn't it.

I've been reading this topic because I have been investigating some sites that look like they are suffering from the same issue.

One site had steady traffic at about 1K per day for months - the performance of the site wasn't great so we optimised the DB and increased site load speed - traffic then went up to 5K per day in steady 500 chunks usually settling at a new level for about 3-4 days before gaining the next chunk. It has levelled out at 5K per day and has been steady for some weeks now.

The other one we deal with has steady traffic with what I would call a normal weekly profile for the industry - except for the fact that it stays at the same maximum level. You can add large sections of content, remove large sections of content, add links, optimise pages all to no avail.

I was fairly sure that improving server speed was going to help site two until Rumbas posted, now I'm not too sure. Rumbas did you increase the pipe to the server as well as increasing the server speed? - Is there any chance that the pipe may be pushing too much traffic through it and slowing down the performance? Have you tried delivering the pages with compression?

From what I am seeing I think there is more than one throttle filter in place. I would suggest that at least two are in play:

No 1. is to do with raw site performance as we have seen evidence that improving performance can break a traffic cap (whether this is Google using its spider as a response time checker or whether it is user response data being filtered back into the algorithm - i.e. site quality goes down when performace is slow - I don't know, or perhaps both come into play.)

No.2 is some kind of traffic equalisation which drall and rumbas seem to be seing, based on rotating SERPs the reasoning behind which I can't fathom at the moment.

That is one possibility - it seems outrageous, but so does this whole throttling idea in the first place, doesn't it.

Maybe Google has set a traffic limit for every domain as a failsafe mechanism to guard against the possibility that some outside agency might penetrate the ranking process, or find a way to manipulate it, so as to produce a huge flood of traffic to one site. The limits could be set high enough that normal traffic to most sites wouldn't be affected.

Good questions Ian.

>did you increase the pipe to the server as well as increasing the server speed?

I moved the complete site, db, dns etc. to a whole new host.
Much faster, more ram, optimized to perform a lot faster with caching, compression, cdn etc.

Load time significantly reduced, fast and slick. It's a heavy site with quite a bit of images - all on cdn and cached.

However GWC still showing under average and very slow times, even after the move to the new server about 4-5 monts ago.

It's actually bugging me as there's no doubt that the users are experiencing a faster site.

The site hasn't move an inch in total search referrals. Ups and downs, but it's a flat line over 1-2 years.

Reno - I think "absented". If it was just a case of +11 or +21 there would surely be at least a ripple on the flat graph from people who really look at SERPS, trying to find a good result and finding it?

+51 I could go with. :)

All of the sudden we have a Y that ranks for a major traffic phrase due to social networking, viral, IBL...

Lets say X's usually shows at position 2-10 for serps.

After Y has spiked all X's get depressed to positions 8-25 for the remainder of that day to balance out the difference of what Y spiked for.

I can confirm that I have definitely seen something very similar, but I had not related it to traffic throttling. The cycling was also slower than what's reported above. Specifically, I noted this in Jan 2009, when there was a lot of concern, as I remember, that Google was "punishing" webmasters for trying to manipulate rankings, and you could never be sure whether, say, acquiring a keyword rich link would bring you up in the serps, or, paradoxically, push you down. I think the discussion was in relation to OOP penalties and to the -950.

I remember that I was seeing a site's apparently unstable reaction to a strong inbound link to home; the link had "good" anchor text, but that text was more relevant to an inner page. I compared the ranking behavior to the behavior of a balloon, where squeezing the balloon in one area would cause another spot to bulge. Here, as the phrase rankings apparently got sqeezed down for the home page, they would bulge up on the inner page.

To use another metaphor, watching the rankings on these pages over time was almost like looking at a seesaw. I didn't think of it as traffic throttling... more as a cyclic sitewide redistribution and resettling of ranking factors.

Or no traffic at all after 1 pm, day after day - even though there are 100s of visits per hour until 1 pm.

This is precisely the time-related kind of query I alluded to earlier in this thread... IMO, most definitely related to work hours, and a completely different effect from the above. I could cite other likely hours as well, but I'm under a non-disclosure not to. I assume this time-related effect does change in time zone by time zone across the country, and, I'll bet, in relevant languages around the world. I would also guess there's no change in rankings associated with this... just a change in the number of searches or bookmark clicks or whatever.

Tedster:

The test is to methodically check position for a small handful of keyword phrases at regular intervals

I think some proxy checking might also be in order to see geographic variations. Or, even better than a proxy (whose IP is usually well known to Google) would be if you have access to a VPN and can use that to check.

You can easily check SERPs position geographically using Adwords.

You need to do the following:

Go to

[adwords.google.com...]

Then enter your keyword and select your location. For every country, they is a sub list of local datacentres.

Then press preview ads, and it brings up the organic SERPs for that location.

It's an eyeopener - you can see exactly where you rank in a whole host of locations.

HTH

Thanks for that tip, Alyssa - I had forgotten about it.

the performance of the site wasn't great so we optimised the DB and increased site load speed - traffic then went up to 5K per day in steady 500 chunks usually settling at a new level for about 3-4 days before gaining the next chunk. It has levelled out at 5K per day and has been steady for some weeks now.

I have seen sites with optimal loading times generate unlimited traffic very quickly after the remedy.

The same sites with poorly optimised DB's applied on otherwise very fast servers crashed with no recovery to a fraction of the original visitations and capped within a set range .

The effects of this are two fold , usability and indexing. Both can and will interact with Google's algo to affect rankings in some way. The correlation between available link juice on a site and it's traffic may well be part of this.

Not all throttling may show this , but this is my view of what's happening to some of this issue.

Can anyone else relate speed and indexing/link juice to this ?

Last year I saw the analytics for one site that showed a drop to zero Google traffic every day at around the same time - it stayed at zero until the next day and then it cycled again. I also saw another major enterprise site that only had first page ranking for a specific 4-hour period every day. Both of the cases persisted for many weeks, but both eventually "returned to normal".

Everything is humming along normally on Monday, all X's (thousands)are getting exactly what they usually gets traffic wise so far into that day.

All of the sudden we have a Y that ranks for a major traffic phrase due to social networking, viral, IBL...

Lets say X's usually shows at position 2-10 for serps.

After Y has spiked all X's get depressed to positions 8-25 for the remainder of that day to balance out the difference of what Y spiked for.

If Y continues to spike and stays at that new level of refers then all X's stay at those new depressed rankings forever.

That's basically what is happening.

Hate to appear out of the blue and debunk this theory, but....

These examples are NOTHING ALIKE!

To IMPLY cause and effect, the examples would have to be "somewhat" similar.

ie. Drall's site would have to get 36,000 hits by 9pm, 5pm, 3pm, etc and then get NO hits for the rest of the day for this analysis to be even CLOSE to a SYSTEMATIC example of INTENTIONAL throttling.

If people want to discuss traffic throttling in terms of different SERPs being shown at different times of day, no argument here.
I will agree to that.

But the 3 (since I only trust Tedster's and Drall's anecdotal analysis...sry everyone else) out of Billions(?) of sites that DO NOT fit the ORIGINAL case that tedster based this theory upon, --
It makes this an ANOMALY, not a consistent theory.

Since NO one has reported a replicant of the FIRST case, the theory is flawed to begin with.
:(

I'd suggest throwing out the FIRST case completely.

Start with Drall's example and only use SIMILAR cases that existed BEFORE 2010.

As unfortunately, not enough webmasters have a thorough knowledge of what happened with Caffeine and esp. during the last few months to report "reliable" data that would support this theory.

It's actually bugging me as there's no doubt that the users are experiencing a faster site.

This is what bugs me with the second site I am working with on this issue - the users get a fast site because of the way the site is designed.

I am currently looking at the firewall speed to see if that makes a difference - a good load testing service would be useful here in identifying whether there is a specific level of traffic that causes a significant drop off in performance.