Zombie Traffic from Google and Traffic Shaping/Throttling - Analysis

[tedster]

We need a dedicated thread to look at this odd phenomenon being reported by a subset of our members. It really isn't about any particular "update" because the apparent signs have been reported since 2008.

I have personally seen just a few examples of traffic shaping and nothing I could really call zombie traffic, but I think it's time for all of us to take the reports seriously and at least give advice on how to analyze what these webmasters are seeing.

To truly make sense of this, we'll need to pull in many areas of Google that we rarely talk about. This ain't your daddy's SEO! Here's a pretty good overview, from 2010: [webmasterworld.com...]

[backdraft7]

Alert: ON period seems to have started 11:30am CST (marked by two rapid succession conversions...it will take a few more to demonstrate a true ON period - update, a third "fast" conversion within 5 minutes, it's ON!). It started off with a converting iPad visitor! There goes my mobile theory! He had a 768x1024 screen resolution though, so that helps. This seems more like fishing every day.
Anyone report similar luck?

[SevenCubed]

There goes my mobile theory!

I would consider an iPad a mobile device too, not by actual definition of what a mobile device usually is, but in the sense that it is readily in use while away from home.

[backdraft7]

Guess that ON period lasted less than 30 minutes. That's typical though.

[tedster]

I've been thinking about the on-and-off traffic question. No conclusions yet, but some ideas/clues.

What I've come up with is load balancing issues. This could be a Google load-balancing switch or it might even be major ISPs doing some kind of load balancing.

I'm not sure about that second bit of technology, but I do know that my home computer sometimes gets identified as being 100 miles away from where I actually live - and yet other times, the location picked is exactly right.

I'm not talking about Google's geo-location either, but various other websites that I visit. It could even be some incorrect data from the IP/location database services. Somehow I think that last one would be pretty short term, however.

[scooterdude]

Is it so difficult to conceive that one corporation might decide to exercise its power to control the incomes of others for ends of its own?

In this circumstance, perhaps some are fed more regularly than others, any sites pop into your mind?

Another question :
How balanced can your view on any organisation be if you refuse to contemplate any possibilities that have negative connotations?

[backdraft7]

By noon today it all shut down again. Nothing now for a solid 5 hours. I've said it before, I've never seen patterns like this, it used to be "natural" spans between conversions, nowadays it is very un-natural / artificial feeling spans. During the off period, mostly one page visits, 0-10seconds. Many Safari visits, outpacing IE 155 to 133 so far today. Chrome 58, FF 41. 31% mobile device users today.

[edited by: backdraft7 at 10:55 pm (utc) on Sep 22, 2012]

[tedster]

Is it so difficult to conceive that one corporation might decide to exercise its power to control the incomes of others for ends of its own?

Of course not - I certainly remember exactly such maneuvers by various companies in years past.

At the same time, I think that for the past year or so, many website owners who are struggling with Google traffic drops immediately jump to the darkest of conclusions. And that rules out any chance of taking effective action to get a decent SEO outcome. Rigorous analysis is a key factor in good SEO.

For example, take this "zombie traffic" phenomenon that a relatively small slice of sites are noticing. After three years, in the just the past few days there is some new analysis that points into a possible direction that WOULD be within the website's control.

If it is something like mobile users being sent to a mobile-unfriendly site, it is baffling that Google hasn't learned something about not continuing to do that. They would not be doing themselves any favors by treating their own search users that way.

I'm more suspicious of the other half of this thread's topic - the "traffic throttling" idea, really almost a traffic quota. I know that Google engineers recently said it's not the case - and yet how do you explain Google traffic shutting off to zero every day between 10:30 am and 1:00 pm local time, always when a certain threshold is reached each day? And I've seen exactly those analytics. And search traffic always started up again at local midnight the next day.

I'm still considering alternate, non-malicious explanations for throttling. With so many automated, machine logic process in play at Google, all kinds of strangeness could emerge that is not intended. And in one case I know of, the phenomenon ended after about 5 months. So something was going on and then it stopped. Reasons not understood - but the return of traffic was a happy thing.

I wish I had had full admin control of that analytics set-up rather than just the permission to take a glimpse. That kind of thing often responds well to some tailored segmentation, or even raw server log analysis.

[backdraft7]

And search traffic always started up again at local midnight the next day.

- a very common pattern that I see lately - a sale right after midnight, one at 3am and one at 7am, then dead all day.

It used to be that 10:30am to 1pm was a HUGE spike for us from the lunchtime crowd.
Now nothing at noon.

Clamping of conversions per day is also common. Variation of +/- 1 or 2 all week. Still trying to get a handle on that phenomenon.

The free flow of traffic that we enjoyed for years now appears to be restricted.

I myself am trying to look outside Google for an explanation.

[TypicalSurfer]

Google may be offering more helpful results to users while sending zombie/zilch traffic to less helpul organic results. :0

In his argument, Cutts said, “We actually think our ads can be as helpful as the search results in some cases. And no, that’s not a new attitude.”

[duckduckgo.com...]

I think that for the past year or so, many website owners who are struggling with Google traffic drops immediately jump to the darkest of conclusions.

Not quite sure google would consider such a thing as being the dark side seeing as their PR puppet is now selling the concept.

[xcoder]

The free flow of traffic that we enjoyed for years now appears to be restricted.

Borderline "criminally" restricted.... I so had enough with this BS that i actually filed a complaint with the ACCC this weekend. Restriction of trade is a very serious offense here (AUS)... let the fun begin...

[gehrlekrona]

@typicalSurfer, I read the same thing kinda' somewhere else where GOG said that if they think their ads are more targetted to the search then they show them. Si a light bulb went off when I saw your post.
I have always been looking at my "key words/phrases" and for the longest time they ssem fine. They are there but people doesn't click. GOOGs ad income has increased a LOT so is it that searchers just click on the ads instead of the SERPs? It could very well be. So in this case, should we be paid by GOOG for using our work that they exploit to make money off of their ads? Then again, we don't even know if GOOG is sending "real people" or just zombies anymore, do we?

[TypicalSurfer]

@gehrlekrona

I've been aware of the obvious correlation between google "updates" and quarterly earnings reports going back to what was referred to as the "Florida update", to ignore such an obvious signal when trying to get your head wrapped around google is a bit lacking in rudimentary analysis.

Truth be told, you can trade google stock (and make easy money) based on how much noise comes out of webmaster forums after an "update", legal frontrunning so to speak. I've yet to see an "update" that didn't significantly enhance revenue for google.

[gehrlekrona]

@typicalsurfer, I know I am a bit slow :) I have been looking at it from a personal view and didn't understand why, even though the site was in the results people didn't click on the links. It was as though GOOG showed different results to anybody. They used to have the same results for a period of time and it didn't really change but now it seems to change evry time you search. One thing I still have to figure out and that is why I get most hits from the east coast......
Maybe we can start a "GOOG buying stock" group and start making some real money instead of hoping GOOG will send us traffic? I'm glad they didn't choose "GOG" as their stock symbol ;)

[bluntforce]

@tedster

Your mention is the only one I believe I've seen with a defined throttling pattern. If an algorithm was producing a throttling effect there should be many instances with the same or a similar pattern.

This does not appear similar to the original sandbox, which existed even though Google did not believe it existed. Here, I haven't seen enough remarks to make me think it is a major facet of the algorithm.

Manual action? Experiment?
I'd lean towards experiment as Google provides significant free time for employees to work on things that interest them.

[tedster]

Your mention is the only one I believe I've seen with a defined throttling pattern.

Agreed - and I've been looking closely for several years to find another. That's one reason why I started this thread. Thinking about it over the past 3 years or so, the suspicious thing for me is that the Google traffic went to zero for that site every day, not just some low level.

There may well have been some other explanation that we didn't uncover - even a hacked server that the owner didn't detect. One rather new hack at the time was grabbing just part of the Google traffic sent to a site - and criminal black hats were testing all kinds of devious hacks, including exploits of DNS servers that aren't configured properly.

So one case of "throttling" is not a likely proof that Google did it. In fact, it's more likely to be an indicator that there was some other explanation.

---------

With "zombie traffic", a solid understanding of what technical mechanism is involved seems to me to be essential as well, before we can say Q.E.D. on any explanation or suspicion.

[rasputinsghost]

Hi all
I'm new to Adsense -got into it by "accident" when a client abandoned a domain that I had developed for him - so I took it over and added some monetizing. - Anyway...

Webwork
I was also wondering if "this system" could be used "to tweak" Google's content network PPC pricing algo/smartpricing". If a well structured bot could affect G's PPC, by artificially lowering/raising bids on the content network, then I imagine that ability could be financially exploited.

I think that Webworks has spotted one of the many components that make up this issue.
When I added an Adsense advert code to a page, I immediately noticed that each time I browsed to the page, there was an immediate "retrieval" of the same page - by a Google related spider/bot. (Think it even had "AdSense" in its name)
And if I remember correctly, when I added a second advert to the page, every retrieval triggered 2 "visits".

So, being an Adsense newbie, I thought that that was the way it should be and carried on.
Did not worry about my stats being skewed because my site engine's statistics module was identifying the visits as being from a spider/bot and the visits were not skewing my "human visitor statistics".

---

But as I mentioned earlier, this is only one component that makes up this issue.

The second component is all the bounce visits.
Could (probably has) become very serious and is probably behind so many guys being thrown off of Adsense recently.
I am sure that everyone has at some time or other been asked by Google to verify your humanity because they have detected that you may in fact be an automated process.
So you type in the captcha and off you go.

Now implementing this on a site may sound simple but there are a whole bunch of things to consider.
1) You do not want to screw around with legitimate spiders/bots that you want to visit your site.
2) You do not want to screw around with legitimate human visitors unnecessarily.
(I hear that Google will not be supporting IE 8 and lower from sometime soon - AT LAST!)
3) You want to detect and store as much information as possible, ESPECIALLY on the bounces, so that you have it available for submission to ISP's, Google (maybe inform them of the nonsense going on) or possibly the legal team of a class action should things need to be taken that far.
4) You may want to make the whole process intelligent and detect if more than 1 bounce comes from a specific ip in the last X seconds and if there has been, ban that ip for 24 hours.
5) There are also certain pages that the "kiddie-hackers" always test for - wordpress admin pages, drupal admin pages, phpbb posting pages and all the rest - so, if you build up a list of "kiddie-hacker" pages that are not on your site (if you are NOT running a wordpress site, anyone who probes for the wordpress admin page is only up to no good) and if they are looking for one of the pages in the list, ban the ip for 24 hours right there and then.

Hows this for an idea?
When an illegal probe is detected, fire of an email immediately to the ISP giving them all the details.
They can then "take up the matter" with their client and you have proof of what has happened.
Imagine some little idiot sitting at his pc up to no good and the phone rings - the ISP phones up and tells him to stop his crap - should make an impression!

Draconian? Definitely - but we are fighting for our lives here.

[Martin Ice Web]

@backdraft,

as we were in the same situation as you ( we sticky mailed ) the last two updates were the first time positive for our site.
I don´t know which one were bringing more traffic but i think it was the anti-domain spamming update. We gained a lot keys but this are not our old keys.
Instead of seeing zombies etc. we now have normal "traffic" with periods of 10-20 min without traffic every 2-3 hours.
Site visits per visitors are up, time on site is up, conversions are up.
I think we are now at prior panda 24 juli levels ( + 50% )
I think that there are two different alogrithm changing back and forth as the quality of the serps are sometimes ok and then awful.

[bluntforce]

One rather new hack at the time was grabbing just part of the Google traffic sent to a site

I'd honestly forgotten all about that. It's probably why I developed the habit of checking my visitor traffic from each major engine against the others.

[backdraft7]

Another "ON" period is happening now. Been "ON" since 6am Saturday to at least 11am. Hope it lasts all day.
I think I'll finally leave my office and enjoy Octoberfest in Appleton, WI today.

[backdraft7]

Kicking the topic again...
The above mentioned "ON" period lasted 12 days and suddenly, FNFR, ended on Thursday 10/11/2012.

It's now Sunday 10/14/2012 and even though my biggest competitor's site finally got hacked (he never upgrades his scripts) traffic and conversions are still sub 2005 levels. I did have a short open period late last night, with 3 US sales while 99% of the nation was sleeping. Now that 99% are awake (presumably) traffic and sales are again shut down. Serps look pretty much unchanged from what they were the past 12 days, yet traffic is nearly shut off. Weird wild stuff.

[tedster]

I have a big request for this thread. Everyone, please use the phrase "Google traffic" and not just "traffic." Some of the comments sound like they are talking about ALL traffic - and that is not a topic for a Google forum. We'll never see anything new if we aren't clearly focused on traffic from Google Search.

I know many of you are just assuming the word "Google" in your posts abou traffic, but many readers are not - and that is adding to confusion. So I've aslo added the word "Google" to this thread's title.

Thanks all for hyour cooperation. Let's not just report on the traffic shifts, let's get to the bottom of this thing. And that begins with being sure that is really IS a Google issue.

[TypicalSurfer]

"zombie traffic" is a bit ambiguous as well. My understanding is that the issue revolves around non-converting G traffic vs. converting?

G has all that juicy converting data from GA and AW, geographical, prime time, seasonal, etc.

[tedster]

Another question comes to mind for those who are reporting zombie traffic. In addition to not converting, what do the other stats look like - page views per visit, time on page, etc. Any big differences from "normal" traffic?

[backdraft7]

@tedster - the vast majority of my traffic is from Google, so yes, Google traffic. As far as what's different? It seems referrals from key sites suddenly stop and strangely my mobile percentage goes down on "bad" days / periods. Other than that, GOOGLE traffic volume is the same on good days vs. bad.
TS - Yes it is ambiguous. It just doesn't shake out clearly in our data.

[diberry]

Backdraft7, in Piwik can you identify a zombie visitor and then take a look at a profile of where they went on your site? My stats program (Clicky) gives visitors a cookie, so it can tell you whether they clicked your page and left after ten seconds, or clicked your page and then clicked a link out or went to another page. It could be helpful to see whether your zombies are just bouncing out or are looking around but not buying.

[edited by: diberry at 5:53 pm (utc) on Oct 16, 2012]

[gehrlekrona]

Not sure this had anything to ho with anything in here but here is a thing I noticed when doing a search today.
Widget in Va gives one result set
Widget for sale in VA gives another result set
Widget in VA for sale gives a totally different set.
Why are the last 2 so different?

[bluntforce]

I don't know that anyone has an exact answer.
1. Widget in VA (Hopefully your search was caps)
2. Widget for sale in VA
3. Widget in VA for sale
If the stop words are removed:
1. Widget VA
2. Widget sale VA
3. Widget VA sale
2 + 3 should provide similar results
There is a difference between "Widget for sale" and "VA for sale" and there is a difference between "Widget for sale" and "Widget in VA".
It appears the search isn't based solely on the keywords without stop words, but filtered possibly by user intent and/or location.

[backdraft7]

@diberry - it's so hard to tell. right now things are moving along fine, I'm seeing many visitors hitting a dozen pages or more and spending quite a bit of time on the site. Conversions are "normal". Suddenly though, this can turn and I will start seeing a series of 1 page visits. I'm in an apparent "ON" period right now. Mobile or non mobile doesn't seem to matter much.

This is beyond any natural patterns I've observed in the past, but you all knew that.

[gadget26]

@backdraft7, I've noticed over the past months that your on/off periods are the exact *OPPOSITE* of mine. It's uncanny! Even the extended periods. Conversions from Google organic traffic have been unusually slow for me since Saturday (10/13). I've started to post about this connection several times, but each time I start I think "nah, that's just crazy." But the coincidence is getting to much to believe.

Do you think we might be on opposite sides of the "your turn" algorithm?

Note that it was harder to tell when the turns were very short (less than a day), but lately the zombie on/off periods seem to be several days long.

By the way, I've driven myself silly trying to figure out what makes a zombie. All I know is that the bounce rate, time on site, and actions per visit are usually all worse when the zombies are hitting me. I have piwik and they look like real people, normal distribution of location and equipment as far as I can tell. The keywords tend to be more of the "informational" rather than "buy" type though. And they tend to bounce after 1 or 2 pages.

Google has far more information available about these searchers than we do and I suspect it is capable of making a pretty fair evaluation (based on previous actions, etc.) about how "hot" these leads are. It could meter them out to various merchants, sending the hot ones one direction for a while, etc. But why? And why alternate over a multi-day period?

It sure makes it hard to run a business smoothly.

Cheers.

[tedster]

The keywords tend to be more of the "informational" rather than "buy" type though.

That may be a very important observation! Here's a theory that immediately popped up for me:

Google has automated taxonomy assignments for query types and for URLs. If the taxonomy assignment algorithm can't figure out where your page "fits", they might keep trying different categories for it in an automated, machine learning fashion.