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...]

[bluntforce]

@Sgt_Kickaxe
Removed affiliate links from higher trafficked pages - signups dropped - rankings did not change. Replaced links today - waiting to see if there's any effect.

@backdraft7
Can't help you due to above test. There's a link in this thread going to a late 2010 discussion about possible Google A/B type (or more complex) testing, it's worth a few re-readings.

[Panthro]

This may be a stretch, but does anyone think this could be some kind of a clue? (from the recent "unnatural links" warning). My emphasis added.

"If you received a message yesterday about unnatural links to your site, don’t panic. In the past, these messages were sent when we took action on a site as a whole. Yesterday, we took another step towards more transparency and began sending messages when we distrust some individual links to a site. While it’s possible for this to indicate potential spammy activity by the site, it can also have innocent reasons. For example, we may take this kind of targeted action to distrust hacked links pointing to an innocent site. The innocent site will get the message as we move towards more transparency, but it’s not necessarily something that you automatically need to worry about.

If we've taken more severe action on your site, you’ll likely notice a drop in search traffic, which you can see in the “Search queries” feature Webmaster Tools for example. As always, if you believe you have been affected by a manual spam action and your site no longer violates the Webmaster Guidelines, go ahead and file a reconsideration request. It’ll take some time for us to process the request, but you will receive a followup message confirming when we’ve processed it."

Some of the phenomena reported here seems to suggest that Google is able to limit traffic without a visible change in SERP rank. Maybe if you're getting throttled, it's some kind of a "more severe action".

Again, this is just a thought but thought I'd put it out there.

[diberry]

Panthro, there are two phenomena people have noticed:

--Traffic Throttling: where Google just seems to stop sending searchers on a dime at some point in each hour/day/week.
--Zombie traffic: the searchers keep coming, but suddenly none of them convert. It's like you're getting a whole different, bad quality of traffic.

Your theory might explain the first IF everyone feeling it has had a penalty. I feel I'm being throttled, and I got the "no manual penalty" letter, but others can chime in. (After all, I may be wrong about my throttling - my searches are so low, it's hard to get reliable statistics from them.)

But I don't think your theory would apply to the second, and that's the one we're mainly talking about in this thread (although there's some overlap, so both versions do come up).

[backdraft7]

Bumping this thread again because after a relatively smooth & normal month of July, they seem to be back.

@diberry - yes, and this time I was on a run for the best week since Penguin hit, then BAM! all conversions stopped at the exact same number of sales as the prior few weeks. It "appears" like throttling by throwing Zombie traffic at the site. Weird! and frustrating!

Also, this time it's just US traffic, not seeing the usually high influx of foreign visitors.

[Vamm]

@Panthro

I wonder how would it be possible to limit traffic with no visible change in search position.

[diberry]

@Backdraft7, sorry to hear that. It's interesting that it's all US traffic though, because I had previously thought the reason the zombies were foreign was because foreign visitors would be less likely to want stuff shipped from the US, or some sites don't even offer international shipping, or they might have local options that are better. I can't recall if you ever said whether any of that might apply to your site, but clearly now we know the zombie pattern doesn't rely on that in any case.

@Vamm, remember we're not all seeing the same serps, they seem to differ from region to region. Maybe it could be done by just shifting them in some areas?

[1script]

I'm seeing a strange thing in my WMT: as soon as zombies appeared ( Thu Aug 2nd) , WMT traffic chart (Search Queries) stopped updating. It's frozen on Aug 1st. For me, anyway. It used to update only two days behind, it's now 5. Anyone else is seeing this?

I was going to check if Google's own CTR also reflects the low conversion quality of the zombie traffic but it does not show the right dates - I keep coming back and it's not updating.

In addition, I'm seeing one site in my account that looks like zombies have started a couple of days earlier: on July 29th the Search Queries chart went up (considerably) and at the same time the clicks went down, and it still stays like that. In other words, looks like the zombies phenomenon is bad for Google as well as webmasters: it would not make much sense for them to show links with significantly lower CTR in their SERPs. Sounds like a waste of valuable screen space.

[tedster]

I'm not clear on what you mean by "zombies" and "conversion." Do you mean people who do not click on your search result when you get a Google impression, or people who do not convert on your site once they arrive on your pages?

[1script]

@tedster: yes to both of your questions.

When we discuss "zombies" here, we usually mean people that come from Google but don't do anything useful on the site. Not necessarily purchasing anything. I have several completely free ways they could otherwise engage with the site, and they don't.

So, my idea was to check whether Google themselves have the same thing happening to them - people search for something but do not click (do not engage) - during the same time it happens to my site.

I don't actually know if checking Google's reported CTR is a valid way of judging the engagement of visitors with Google - it could be simply a function of them dropping my sites down a few position in SERPs. However, if the overall Google traffic holds steady (and it is so far), then seeing lower Google CTR would mean that they started to show my site to a larger audience but it is a wrong audience (hence zombies). Wrong for both me and Google - I've got to think that we're on the same side here.

I hope I'm getting this right and not just confusing things in my head, please correct me if I'm wrong.

But, regardless, I'm not able to test that hypothesis anyway: my zombies appeared on the 2nd of August but my last updated Google WMT Search Queries are August 1st. This 5-days delay with update is rather unusual, that's the info I wanted to get out to check if other people are seeing this.

[diberry]

I keep trying to think of a group of humans who could account for these "zombies" (as defined by not engaging with one's site meaningfully). Spammers from particular time zones? Nope, backdraft7 is getting US based zombies, and there's no reason US spammers would be showing up without normal US visitors.

It has to have something to do with Google's geo-location stuff, I think, and/or testing different data sets. And yet, 1script's account of the possibility that the zombie traffic is also doing Google little good puts a new light on that.

So, here's a question I don't think anyone's asked? What if these visitors weren't people? What if they were bots or something like that? Can anyone imagine any sort of scheme along those lines that would benefit anyone, anywhere? I can't, but it's a possibility, so I thought I'd raise it. (It still wouldn't explain where the more typical visitors are disappearing to during those times.)

[1script]

What if these visitors weren't people? What if they were bots or something like that?

I thought of that and I do, in fact, catch quite a few rogue robots trying to appear as if they came from Google on very bizarre keywords, something I would never be ranking on and can't find my site anywhere on that SERP when checked. But the amount is not all that significant compared to the normal Google traffic. Also, some of the competitors are checking up on me automatically and their software appears to have google.com setup as the referrer.

So, I guess, it is a valid idea but it cannot account for the significant difference in engagement. Besides, for the total traffic to be stable, the robot attack would have to be timed perfectly with the dip in real Google traffic - I don't see even a remote possibility of that.

I can't see how anyone would benefit from that either. Perhaps I'm just not feeling particularly imaginative today ...

By the way: the Search Queries report in WMT hasn't been updated in 6 days now - this is very peculiar. Calm before a storm?

[ohno]

By the way: the Search Queries report in WMT hasn't been updated in 6 days now - this is very peculiar. Calm before a storm?

Same here but not sure I trust that data anyway, I've got a screen shot of that graph showing a NEGATIVE click! How on earth can you have a -1 click?! Is there a way to post the screen shot on here?

[lucy24]

By the way: the Search Queries report in WMT hasn't been updated in 6 days now - this is very peculiar. Calm before a storm?

Me too-- and it's not like that would be a major drain on their computing resources.

Has the change rate gone berserk? I don't normally look at it since they took it off the default page, but the table is simply splattered with huge numbers, bottoming out at -100%. (Uh.... They're listing all the search terms that people liked 30 days ago, and will never find again, just to show that google giveth and google taketh away?) Position changes look as if they just made it up and gambled nobody would never look, because I really doubt that all those terms have gone up or down by exact multiples of 100 or 10.

[1script]

By the way: the Search Queries report in WMT hasn't been updated in 6 days now - this is very peculiar. Calm before a storm?

Well, there has been some change here. In the last 12 hours the report had advanced 2 days for some sites and 1 day for others, which in itself is very interesting: I've never seen the results to become dis-synchronized across sites in the same WMT account.

Back to the main subject matter: I can only see one "zombie day" so far (Aug 2, 2012) but it shows that Google's own CTR is also lower. I know one day is not much statistically, but if the CTR holds, perhaps it is a hint that one of Google's latest algo changes has been detrimental for relevancy of Google's results. That would definitely result in decreased CTR across the board.

Actually, even the advertisers would be affected - bad traffic is bad for everyone. I noticed that on sites that run ads CPC and CPM went down sharply (~20%) on Aug 2nd. Apparently, bad traffic converts badly in every sense of the word "conversion" and it looks like those advertisers that watch trends very closely have noticed it right away. I would trust them - it's their money on the line...

[backdraft7]

So, yesterday, had two conversions, one at 8:16am and one at 8:38am...so it's looks like a good start to the day...NOT! The next conversion occurred at 5:01pm and nothing else for the day.
Then, tonight , I get conversions at 12:23am, 1:10am and 2:36am.

Is Google throttling my site? Absolutely!

Is it intentional? Probably not, but it just goes to show how dynamically screwed up their search engine has become. My typical conversion pattern (for years) was one on the hour plus or minus one.

I really need a new strategy for tracking traffic better because I'd love to get to the bottom of this issue.

[backdraft7]

Gonna have to call it again...higher traffic levels, lower quality, no conversion, happening now.
bumps

[ohno]

Same here. Next to dead then a spurt of non converting Zombie junk. This time though the periods between are much longer, high percentage of just nothing at all. Whatever they started last Friday has well & truly killed us. Job done Google?

[timwilliams]

Unique day so far today, organic traffic conversion rate up (almost double), adwords conversion rate normal.

[backdraft7]

Getting a load of Zombies again on Monday, Labor Day. 4 conversions all day compared to 16 last Labor Day.
Plenty of traffic, but bounce rate has shot up 5% all by itself with this traffic. Obviously non targeted Zombies.

[backdraft7]

Going on a second consecutive day of higher volume, but definitely lower quality traffic.

[Martin Ice Web]

It has to have something to do with Google's geo-location stuff

I read somewhere that google says 90% of the queries are local related?! I doubt this. 99% of my queries are not local related.

We got a good day yesterday, traffic was up better then normal. But - and I did know it - today the opposite. Very, very low traffic overnight. So the overall traffic ( night + day ) stays the same.

[Robert Charlton]

I read somewhere that google says 90% of the queries are local related?! I doubt this. 99% of my queries are not local related.

Google's figure, which I've seen repeated numerous times, is that 20% of all searches are local related.

This is a percentage of "all" searches. It's not about you. It is about arithmetic. If we look at 5 sites, all with equal traffic, and 4 had absolutely no local traffic, and one was entirely local, we'd say that 20% of the traffic for those sites was local.

How local relates to this discussion

Much speculation to follow.... "Zombie traffic" appears to be traffic from locations that don't convert. It's been observed among people who monitor Google search rankings closely that rankings are skewed locally. "Hyper-local" was the term used by Greg Boser at last year's PubCon. Ditto, as has been observed in various threads on zombie traffic, throttling, and shaping, Google auto-complete suggestions appear to be skewed locally. So, localization may be a useful framework in which Google can implant results for whatever it's testing.

These test factors may or may not include geo-location.

I've also noted some results that aren't local on the surface might be affected by local factors. In this discussion, I noticed search results shifting as I changed my default location by only several miles closer to a retail outlet for a chain that was ranking for a particular search on which we allowed discussion.

Bullet Points in SERP description, is that new?
http://www.webmasterworld.com/google/4355288.htm [webmasterworld.com]

I noted...

I mention all this because I've felt that all those extra organic results we've been seeing, the multiple results and the Local listings, including those with Place flags, have been part of ongoing tests, changing with each iteration of Panda (or whatever else it is that Google might be testing).

So, that's in part what the "local" talk is about.

Does local still relate to this discussion?

The bullet-point test noted above may or may not have been a test of what I'll call QDL... Query Deserves Localization.

Let me toss out this thought, though, that some local results may have stabilized and Google may be into testing other factors which don't related to locality at all. It's possible, therefore, that the zombie traffic some are getting may no longer be locally related... that it might also or instead be, say, Panda related. For me, this is speculation, and I haven't observed zombie traffic directly.

I have seen and discussed graphs from other SEOs of what appear to be traffic caps, though... the "buzz saw" effect... which vividly depicts traffic throttling. The only explanation I have for throttling is that Google might be trying to bring, say, various observed user behavior from different sites into the same range for comparison, and limiting traffic for some sites might be a way of doing this.

I'd further imagine that Google might do this only in marginal areas. I can't imagine Google throttling, say, Amazon, at least not for any extended period, in order to bring the stats of all other sites into range.

[backdraft7]

It's the Magical Algorithmic Daily Double Everflux Randomizer. When it's hot it's hot - when it's not it's not.

[Robert Charlton]

Magical Algorithmic Daily Double Everflux Randomizer

LOL, backdraft7.

I can tell you're MADDER than I am about this. ;)

[timwilliams]

MADDER madness, I fully expect [insert big pharma company name] to come up with a medication for this soon! I can totally see the late night infomercial now:
"google got you down? You may be suffering from MADDER madness and we can help*"

_{*If you experience properly converting traffic for more than 4 hours turn off your server and call your Pakistani SEO firm immediately.}

[backdraft7]

;^D

[ohno]

*If you experience properly converting traffic for more than 4 hours turn off your server and call your Pakistani SEO firm immediately.

:D Yesetrday saw a big return to the zombie traffic spurts, zero to 8 visitors in 60 seconds yet not one cart add.

[tedster]

It seems very clear that these zombie or throttling patterns are not timed the same across different sites for different members. The analysis cannot make any progress if all we post is isolated reports like this. So can we please have a bit more analysis in this thread, please? For example:

1. Is there a source country difference in your traffic when traffic levels stay the same but the traffic turns zombie?

2. Are the keywords sending zombie traffic any different than keywords that send converting traffic?

[timwilliams]

Sorry for getting a little sideways there Tedster, sometimes it's all I can do just to laugh about it.

seems very clear that these zombie or throttling patterns are not timed the same across different sites for different members

I don't think that's as clear as it seems to be. I can predict when we are going to see zombie traffic posts based on how my traffic is converting. WebmasterWorld members are a small subset of the internet and WebmasterWorld members that operate direct to consumer ecom sites are a small percentage of the membership (no facts to back that up, just an educated guess).

For me, conversions/sales are the key to this zombie traffic. I watch my conversions like a hawk, constantly trying to make my site more efficient for my customers with the conversion rate being the gauge of that efficiency.

From what I can tell so far, during times of low conversion rates, google is sending the same traffic (same country, same keywords) to the wrong page.

For example: keyword "Green California Widget"

During normal times google displays the following page in the results: example.com/california/widget/green (conversion rate: 7.9%)

During times of low conversion rate google displays a page like example.com/info/montana/red-widgets-new-for-montana.html (conversion rate: 1.7%)

I don't know why they do this, my guess is it has something to do with geolocation, google trying to guess what the searcher wants based on location + search query.

You don't have to try and catch this happening in the serps, you can see this in WMT, (traffic > search queries > query) gives you a list of the urls served in the results for that keyword/query. Some of the keywords I target have many, many different urls that are displayed at any given time.

[tedster]

google is sending the same traffic (same country, same keywords) to the wrong page.

Thank you Tim - that is a useful observation. Does anyone else see this pattern?