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]

Ohno: I think many feel they are selling their soul when they install GA code on their sites. I was ok with it years ago when they were considered very "trustworthy", but since the "don't be evil" statement and the "love letter", not to mention the 'black box lack of transparency' that continues, and the erratic service of late, it's hard to love anymore. I don't hate G, I just don't love or trust them like I used to, and that's too bad. If they read this forum, I hope I'm not slapped for expressing my honest opinion.

If only they'd give us a reason to love them again, but every update seems like another nail. Why is that?

[bluntforce]

I've seen both what appeared to be throttling and the "bunched" signup issues. I believe apparent throttling is just a byproduct of an established website having stable positions for some given key phrases. When I look at Google, Bing, Yahoo and AOL, Google will produce 6-7 times the visitors as Bing and Yahoo, AOL produces 1/3rd as many as Bing/Yahoo. Daily/hourly fluctuations don't mean much but I'll look a little closer if it appears there is a trend. AOL only having significance because of the relationship with Google. I'll see overall search volume drop at various times, but as long as the percentages between the search engines remain consistent it's a user thing, not search related. Average search query on Google is now something over 4 words, I've expanded 2 word phrases making them more diverse and inclusive of transactional type terms.

I've never been able to determine a relationship between "bunched" purchases although I use Piwik, GA, a log analyzer and a database program for in-depth analysis.

[Robert Charlton]

I wanted to note here that Jez123 posted this observation in the Sept 2012 SERPs Update thread [webmasterworld.com...] about 8 hrs ago... msg:4497040...

For some reason my site is doing well in google.au Not bad - only half a world away! Why would I be getting traffic from Australia when my site is UK?

My follow-up post notes similarities between this and what I'd mentioned earlier in this Zombie Traffic thread, suggesting perhaps a test of geo-intent among, eg, former British Empire countries having similar place names.

Here's what I'm guessing....

The gist of it is that there are a bunch of town names that have either identical or very similar spellings, and I believe that some queries get modified by Google and/or sent to a different country because there are no local sites satisfying the query.

This seems to show up the most relatively soon after an update has rolled out. I think this may be Google "calibrating" the results, but that's a guess.

So, is their geo-intent related to your queries, or could there be? (Eg, ecommerce results might get localized simply because Google wants to return local results first).

Anyone else seeing similar anomalies? It might be a clue to some types of zombie traffic. I'm not sure.

[SevenCubed]

...suggesting perhaps a test of geo-intent among, eg, former British Empire countries having similar place names.

In the past when I used google I would run into that frequently for the exact reason of similar place names -- we share a lot of common ones with the UK. This was around years ago so I doubt it is a test as much as just a need to refine a search term by adding a province name which then excluded all the .uk results.

[xcoder]

@backdraft7

Piwik has a Goals tab which you can use to track conversions

It also has a referrer tab where you can track ad campaign clicks, keyowrds etc.

A true state of the art tracking software, comes with a widget board where you can drag and drop a range of various real time reporting widgets and build your own kind of reporting board. It all seems a little complicated at first but once you get used to it you learn to appreciate it... send me a PM if you need any tips.

IP to country translation is slightly off though. A quick plugin update takes care of that...

P.S.
You can have both installed (GA & Piwik)...makes no difference.

[backdraft7]

Wow! what an eye opener!
I installed piwik (and left GA in place) so now I can verify GA figures.

The whole thing now makes pretty good sense.
Zombies are simply mobile users. My site and product is not particularly mobile friendly.

Piwik's 'knuckle dragger easy' live visitor widget makes it so obvious to see.

Solution?
Rework the site to be more mobile friendly.
Easy enough.

I highly recommend piwik!
and many thanks to xcoder!

[timwilliams]

that's interesting information, so when you are seeing zombie mobile users are your regular users lower? What I guess I'm saying is, is there a switch being flipped that sends you only mobile searchers during slow conversion periods?

[backdraft7]

@timwilliams - If I ever get another ON period, I'll tell you.

Update - one sale just came in - it was a desktop user, windows 7 with heavy (50+) page engagement before they bought. I could see that one coming a mile away. It seems every other visit is an iPad, iPhone or android device. Those are usually one page engagement for 0s (zero seconds).

The progression over the past two years makes sense. In talking to some younger friends (early 30's), they say all their friends are dropping their home ISP accounts (due to the economy) and switching their internet access to their smart phone. So, I can easily see how some businesses are thriving in that environment, while others fail. MY site, being older, is not up to the task of serving mobile users, yet. It will shortly, then maybe those 0s visits will increase their stay to become buyers.

Piwik sure is an eye opener. Did I said that already? ;^)

I guess it was obvious, but I didn't want to believe it.
Zombies are mobile users!


I'll try to correlate specific periods, but so far it's pretty obvious.

[xcoder]

@backdraft7

Told ya :)

Now go a head and set goals from the goals tab (click the "goals overview" and scroll all the way the bottom to set a new goal). Now let it run for a couple of weeks and watch the amazing insights... you can thank me later... :)

[tedster]

Wow - if others who have notices zombie traffic see the same thing, this really helps to resolve a significant mystery. Thanks!

[xcoder]

The zombies (in my case) consist of one hit (0 sec) non converting foreign traffic. A quick analysis of their referrer, op systems, browsers, screen sizes, plugins etc suggest lost search souls. I think they're alive and real, just landed on the pages by mistake.

GA does a great job of hiding all this essential information from webmasters...

[ohno]

Same here at present xcoder, sadly no mobile light bulb switched on here. I still think it is time to make sites mobile freindly mind. The other mystery is why Zombies appear all of a sudden? I'm talking zero traffic for hours to 10 visitors within 60 seconds, if that were all mobile users then "they" must be testing something but this has been going on for quite some time....

[edited by: tedster at 7:04 am (utc) on Sep 21, 2012]

[xcoder]

The other mystery is why Zombies appear all of a sudden

Thousands of alternating servers across many countries and one hell of out of whack geo location targeting system. Then add to that mix an out of whack "auto complete" and you get the perfect storm.

I use ebay as a perfect example. It goes all quite, and i mean dead quite whenever some p/p update is being pushed (i have more then 500 product listed there) and a few days later i start getting foreign orders form around the world. I normally target the US & AU markets only, so getting orders from many other countries all of a sudden really stick out...

[xcoder]

Those are usually one page engagement for 0s (zero seconds).

BTW, the "zero seconds" thing does not necessarily mean the viewer really spent only one second. He/she may as well spent 20 minutes reading that very page but clicked back right after that.

Most tracking systems (i know of, including piwik) are able to track the "real" time spent *between page loads* only... so in other words if no other page was loaded on that very same visit then the system will log zero seconds.

[backdraft7]

Yes, We know Zombies are alive, they just don't buy on particular sites, like mine because my product is not easily used on a small screen. If they are not buying on sites that ship & sell tangibles, it's likely because your site navigation is not mobile friendly, text too small, hard to hit hyperlinks, etc.

I think for once we have a handle on the cause & effect. It makes perfect sense.

Time to get out my galaxy tab 10.1 and start testing for better mobile useability.

[claaarky]

"It makes perfect sense."

Certainly rings true here. Our conversion rate for iPad is roughly the same as for desktops, but for smartphones it's terrible.

We also have sections of the site that don't convert well, so if traffic was mostly going there at certain times of the day it would explain the dead periods (traffic remains constant, conversions are dead).

I wonder if mobile friendly sites can bring conversions up to desktop/iPad rates though? Personally I use my smartphone for research and place orders via my laptop because it is a bit fiddly doing it on a phone.

[timwilliams]

Piwik +1

[netmeg]

(not to take anything away from Piwik, which I have no experience with, but you can also find out mobile traffic stats in GA. Just in case you're not thinking to switch)

If this is the cause of the zombies, glad you finally got a handle on solving it.

[claaarky]

I don't think you can see real time mobile traffic stats in GA though, only the total for the day so far or for a specific day (as far as I can see).

If your dead period happens for 4 hours in the middle of the day and that's down to a sudden surge of mobile users I can't think how you'd nail that down in GA.

[Dymero]

@claarky

I believe this can be achieved by creating a custom report filtered by dimension Mobile (Yes) and use Hour for Dimension Drilldown. Then just compare to non-mobile traffic in the same way.

[gehrlekrona]

I'd give anything for some zombies by now....

[claaarky]

Thanks dymero.

[xcoder]

Time to get out my galaxy tab 10.1 and start testing for better mobile useability.

Getting rid of flash objects and any mouseover effects are a good start...

Conversions on small mobile screens sucks. Tablets are OK.

[edited by: xcoder at 10:23 pm (utc) on Sep 21, 2012]

[gehrlekrona]

So now when we have the zombies figured out, how does google start sending zombies that doesn't convert and then switch to "real people"? Do they send mobile traffic for 30 minutes and then desktop people for an hour? Does Google not just retrieve search results and send it to anyone who requests it or is it personalized even down to what phone you have?

[scooterdude]

So I one chap decides his Zombies are mobile surfers and all suddenly decide that the answer is arrived at ?

Begin all your pardons, I wouldn't normally be so insistent, but this thread presents certain opportunities that might be worth pursuing


NO, the zombies might include mobile surfers, but are not necessarily the primary reason for zombie traffic.

[SevenCubed]

So now when we have the zombies figured out, how does google start sending zombies that doesn't convert and then switch to "real people"? Do they send mobile traffic for 30 minutes and then desktop people for an hour? Does Google not just retrieve search results and send it to anyone who requests it or is it personalized even down to what phone you have?

Interesting turn of events here with this zombie stuff if it's leading in the right direction. I'll throw my thoughts into the ring, framed around gehrlekrona comments.

Lets consider it's not google intentionally sending zombies. That implies they are smart enough to differentiate between good searchers and bad searchers but they cannot because they cannot differentiate between good SERPs and bad ones -- something that's under their control -- zombie stuff isn't so it's beyond their intentional control.

What if the packets of zombies are people addicted to the internet? As in, it is periods of transitions when they leave home in the morning (their desktops), reconnect via their mobile devices (while stuck in commuter hell), then reconnect again at work via desktop, then reconnect again via mobile on the homeward bound trip, and finally make their purchase from home PC after having researched it at various times during the day.

Maybe more insight can be dug up by those who monitor and track this zombie stuff by paying attention to time of day it is most prevalent. Keep in mind there are many time zones, especially in North America, and so it can be a sliding scale. For what it may or may not be worth, I'm just thinking out loud.

[diberry]

On a side note, i would pay money if someone would teach me how to REALLY use my stats. I feel like with every package I've used, there's so much more I could be understanding if only I knew how to look at all that data. I mean, maybe the reason it's hard to get good analysis from a lot of us here is that we don't really know how to crunch our stats.

[gehrlekrona]

I am a stats addict and have been watching everything closely for years. I used GA for a short period but didn't want to share my whole life with Big Brother so I removed it. I went to another stat counter package and now I also have Piwik (mentioned in other threads). Looking at the latest package I see that SE's don't always send the keywords anymore in the URL, something I have suspected for quite a while now, so we have less to go after. I think it is something they don't want us to know because then it'll be harder for us to target keywords. As for Google Piwik says that Google doesn't send keywords when you are logged in. Where that is true or not, I don't know, but it makes sense. I am betting they store all that in the user profile instead so if you are logged in in your Google account they know your every move. SCARY!
Zombies? To me it looks like that Google is serving up multiple sets of results on a rotating/random schedule. Why? Because I can't understand why I have certain phrases coming up under, let's say 30 minutes and then other long tail phrases the next 30 minutes. If I look back all that was mixed so you had your shorter keywords AND longtail mixed but not anymore. This might be where the zombies come in too. You have the perfect storm with phone/iPad/Kindle users are searching and Google "happen" to show a set that those users get served?
Back to some stat gazing!

[backdraft7]

I agree with you folks, mobile is NOT the ONLY cause for Zombie traffic, but I think it's a big part.
I do agree that even the PC traffic has been of poor (non converting) quality all this year (Since April).
My thought is that Google is doing something similar to what grocery stores do from time to time...they rearrange the whole store so you CAN'T find anything...that way, in your search, you discover OTHER things you weren't looking for. Ridiculous, but it leads people through many more loopholes and ads than if they find what the need right away. Did I mention Ads?

Just this morning, my mobile traffic is at 36% - wow! 2 or 3 years ago when sales were booming and high speed mobile access not readily available, that might have been 2 or 3 %. At the end of a typical day, it's been around 23%. That's a considerable chunk of traffic.

I'm not so paranoid to think that Google is manipulating anything based on GA usage, but setting aside the mobile issue, there still seems to be a randomness to on off periods. I wish I could track those better, but since the 4th of September, I haven't had any ON periods.

[gehrlekrona]

I too have a lot of mobile users and have had for a while. I created an app for my site that people have downloaded and I have tracked the usage. I agree with @backdraft7 that Google do rearrange the results (see my post above) in that they have lots of different sets of results for the same queries.
One thing I just saw was when I was searching for something for my cat and when the result came up it was kinf of what I was looking for BUT it had a suggestion for me "did you mean...." so I clicked on that and it had a LOT of ads for just a little change of a word/keyword... So yes, the rearrange results depending on how many high(?) paying ads they have. No doubt about it.