Categories: Crawling, indexing & ranking :

Bounce Rate Attack

Showing 1-34 of 34 messages
Bounce Rate Attack Klever Freire 1/9/13 9:23 PM
I've read the FAQs and searched the help center. 

My website has recently seen a dramatic increase in bounce rate.  Our bounce rate is typically between 1% to 3% on a monthly basis.  However, starting in December our website has been attacked (for lack of a better term) by someone trying to artificially increase our bounce rate in an effort to decrease our search engine ranking- and they have succeeded. 

We typically get between 10 and 20 hits per day from natural search results with less than 1% bounce rate.  On days we are attacked we receive two or three times our normal amount of natural search engine results and bounce rate increases to upwards of 40%.  Google analytics shows that all bounced hits to our website come from a source that does not identify its geographical source to be tracked.

Its clear that this is a direct effort to reduce our websites reputation and natural search result ranking.  It seems that Google does not have a way to identify this kind of Fraud.  Our business and web search rankings are suffering as a result!! What can we do!!??

Any help is greatly appreciated!!

Kind regards,

Klever
Re: Bounce Rate Attack seo101 1/9/13 10:01 PM
What has bounce rate got to do with rankings?

If you are paranoid enough to believe its a factor and this is what happened, then remove Google analytics and then how will Google even know what your bounce rate is?

If your rankings have dropped, look for other reasons.
Re: Bounce Rate Attack StevieD_Web 1/9/13 10:02 PM
I think you need to have the firmware for your tinfoil beanie updated.

A)  Bounce rate is but one of well over 200 different factors Google considers when evaluating websites.

B)  You would have to be completely delusional if you thought bounce rate was the most important factor.

C)  NOBODY and I am mean NOBODY has a bounce rate of 1% unless the traffic count is extremely small such that a single event (or non-event) could have dramatic impacts

D)  40% Bounce rate seems pretty darn normal.  That means the visitor came, visited a single page, saw what needed to be seen and returned to Google to complete other tasks.

Re: Bounce Rate Attack StevieD_Web 1/9/13 10:03 PM
oh, yea, what seo101 said..... look for other causes.
Re: Bounce Rate Attack seo101 1/9/13 10:03 PM
..for example, I can see no backlinks to your site which would easily explain why the site is not ranking very well.
Re: Bounce Rate Attack Brian Ussery 1/9/13 10:05 PM
Klever FreireLevel,

Google does not use Google Analytics data for rankings if they did big companies that use other analytics platforms would not rank in Google search results.

Because Google does not use Google Analtyics data for rankings, they do not use Google Analytics bounce rate data either.  Even if Google did use Google Analytics data for rankings, bounce rate would be of little value.  Basically a high bounce rate can mean that users found exactly what they were looking for and then left the site.

Either way, according to http://www.webpagetest.org/result/130110_71_3H2/ the URL provided takes 48.57 seconds to load after Google Analytics is loaded.  Most users will only wait 6 seconds at the most for a page to load.  When analytics loads but users only see a blank screen they think the site is dead and leave but Google Analytics records a bounce.

I recommend monitoring your site and dramatically improving the load time of your imagery.  What you are seeing in analytics as high bounce rate is most likely the result of slow performance.


-Brian

Re: Bounce Rate Attack fedesasso 1/10/13 4:08 AM
Sorry guys, I have to put a word in Klever Freire's defence.
I think there is a misunderstanding in what he said. 

Google never publicly acknowledged as far as I know, but it's commonly accepted it measure bounce rate indirectly as the time it takes to a user to click on a link in the SERP, and then bounce back to try another link (for example, both Bing and Yandex have declared to measure this user behaviour as a ranking signal, as it leads to better results). Bounce Rate as measured by a tool like Google Analytics is in this case just what makes a bell ring.
@Klever Freire, is that what you meant?

If I well understood Klever Freire's "attack" theory, an automated visual agent would simulate a user clicking on his link on the SERP, than clicking on another link (supposedly the one the attacker wish to show G as a better user choice).
@Klever Freire, is that what you meant talking about an "attack"?

That said, I'm not saying I'm endorsing the diagnosis about an "attack" is correct, nor I'm saying such an attack would significantly impact his ranking (I don't know it).

Klever Freire: if your theory of an such an "attack" were true, you should be able to see in your server logs the alleged attacker's requests decorated with an "HTTP Referral" attribute from Google SERP (if referral tracking is enabled).

True, reported CTR are weird, but for a very few visits coming from brand-related queries from google search only, 1% CTR might be real.

Again, I'm not endorsing the "ctr-attack" theory, other causes might the ranking drop.
As beussery pointed out, the home page is really slow to show up something actionable, about 30s.
Re: Bounce Rate Attack Suzanneh 1/10/13 4:29 AM

Your bounce rate is low because you have two instances of Analytics on your page.  (1% to 2% is not a normal bounce rate.)  (Sorry for the bold/red; it's a one-line post and thought it'd get lost in the discussion! :-) )

Suzanne
Re: Bounce Rate Attack Klever Freire 1/10/13 4:43 AM
Hi Fedesasso,

Yes, you understand exactly what I am referring to.  The alleged attacker's requests all have HTTP referral attribute.  I have done a design of experiment statistical analysis specifically with bounce rate as it is correlated to search engine rankings.  If you haven't done this type of statistical analysis on variables that impact search engine rankings I invite you to do so to see for your self the statistical validity of any "SEO 101" claims.  Design of experiments statistical analysis is what is employed when it is not well known how a particular variable effects the results of a physical process or an undefined algorithm- I may be new at some of the web master stuff, but I am not new to science...trust me on this method of analysis and its validity for improving results when information is missing or not available.

The "bounce rate attacks" are happening on highly regular intervals now.  Most days (25-28 days of any month) our bounce rate from natural search engine results is in alignment with our historical trends (~1-3%), but we will then experience one or two days per month where the bounce rate from google search engine referrals goes from 1% to upwards of 44%.  Of course, our hits from google search engine referrals also spike unexpectedly on those days to more than 3 times the normal (historical) values.  As I said previously, the geographic location is also not set for the bounced referrals on those "attack days" only.  

In less than one year our website has become a top 10 search engine result in a highly competitive local market for search terms such as "toronto home staging".  Local search results for our key word target terms are in the range of 2500 searches per month.

I agree that our page is taking far too long to load today, this is a new issue that I am currently working on with the hosting service...

The question is: how to notify Google (and other search engines about this) and eliminate the impact it is having on search engine rankings.

 



Re: Bounce Rate Attack Klever Freire 1/10/13 4:50 AM
Suzanne,

Please explain a bit further.  You point is understood, but I believe it may be incorrect- it doesn't mathematically jive that some days Google would correct itself and some days Google would not based on more than one instance of Analytics.  If your comment is correct, than my bounce rate should just be consistently wrong.  I agree that 1%-2% bounce rate is not typical for high volume websites with a wider range of content types, but we only get hits for very specific search terms in a local market that only sees about 2500 local searches per month for those terms.  When people type in "Toronto Home Staging" they know exactly what to expect...
Re: Bounce Rate Attack Brian Ussery 1/10/13 4:53 AM
fedesasso,

Correct, what you are talking about are "short clicks" and "long clicks" (now publicly confirmed via "In The Plex") but these are different than bounce rate. 

- Short clicks are searches where the user quickly comes back to Google and does another search. (as in results did not align with intent)

- Long clicks are searches where the user doesn't search again immediately. (as in results did align with intent)

Either way, I recommend focusing on speed and analytics implementation issues as the best solution in this case.

-Brian

Re: Bounce Rate Attack fedesasso 1/10/13 5:46 AM
@ beussery/Brian, thank you for the info (and I will by the book).
Are "short clicks" and "long clicks" the official names used in "In The Plex"?
(The term BR was used once by a Bing representative, while the Yandex one called it behavioural analysis or something like that. Sorry, I got no links at hand for reference.)
So, what Klever Freire is referring could be probably named "long clicks attack".

Btw: Reading my own post I realized I used twice the acronym CTR in place of BR (Bounce Rate). I'm fluent with both the concepts, just am so much more used on writing the first one that I wrote it even having BR in mind.

@Klever Freire, I find what you write extremely interesting (I'm a little confused to what traffic segments your BR values are referring to). 
I'm afraid I personally cannot answer your question "how to notify Google..."

Kindest of regards
Re: Bounce Rate Attack Becky Sharpe 1/10/13 5:54 AM
You haven't said just how it is affecting your rankings? Just what is the effect?
Re: Bounce Rate Attack Suzanneh 1/10/13 6:05 AM
It could be any number of reasons why some days show a different bounce rate.  It's irrelevant to the fact that there should only be one instance of Analytics on the page.  Remove one instance and you'll see what your normal bounce rate is.  Even if you don't believe me, believe Google when they say there should only be one instance of Analytics.

Even highly relevant websites will not have a such a low bounce rate.  People are people.  They take one look and bounce away for any number of reasons.

Suzanne
Re: Bounce Rate Attack Suzanneh 1/10/13 6:16 AM
Here's an article with a few reasons why bounce rate can be wrong due to bad Analytics implementation:

http://www.analytics-ninja.com/blog/2012/06/google-analytics-bounce-rate-demystified.html

Loved this part:

Additionally, and I’m still surprised by how many times I see this, having the GA tracking code more than once on a page is a sure fire way to bring your bounce rate down to zero.  As my friend Caleb Whitmore puts it, “A 3.8% bounce rate isn’t really good, it’s broken.”

Suzanne
Klever Freire 1/10/13 6:37 AM <This message has been deleted.>
Re: Bounce Rate Attack Klever Freire 1/10/13 6:38 AM
My GA code is implemented through wordpress.  Wordpress asks for your analytics code in a single place, which is just a copy/paste exercise for the code that Google generates for you.  If you view may page source you will find the following two instances of code related to analytics and they are not the same:

------
One
-------
<script type="text/javascript">

  var _gaq = _gaq || [];
  _gaq.push(['_setAccount', 'UA-18910772-2']);
  _gaq.push(['_trackPageview']);

  (function() {
    var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true;
    ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js';
    var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s);
  })();

</script>

------
Two
------

<script type="text/javascript">
try {
var pageTracker = _gat._getTracker("UA-18910772-2");
pageTracker._trackPageview();
} catch(err) {}</script>

--

I don't believe that my Google analytics is implemented incorrectly, but please explain to me how its incorrect if you believe it is.  If it is am incorrect implementation then I will have to troubleshoot how wordpress auto-incorporates the analytics code.
Re: Bounce Rate Attack Suzanneh 1/10/13 6:47 AM
No, it's not implemented correctly..  It doesn't matter that you only enter it in one place in WordPress, something else -- either in your template or in another plugin --- is duplicating the code.

See the trackPageView()?  You have that in there twice.  TWO times. So, Analytics goes:  page view, page view.  Voila.  Low bounce rate because it believes the visitor viewed two pages and, therefore, did not bounce.

You have two choices:

1) Remove the extra instance and have real stats for your site; ask people for real reasons why you may have had a drop in search results.
2) Continue to believe you have the best site around and that any drop in traffic is due to a bounce attack.

Suzanne
Re: Bounce Rate Attack Suzanneh 1/10/13 6:51 AM
This is what you need to get rid of:


------
Two
------

<script type="text/javascript">
try {
var pageTracker = _gat._getTracker("UA-18910772-2");
pageTracker._trackPageview();
} catch(err) {}</script>

The script to run that code is already in the first instance.    Now you show me where Google says you should have it twice:  https://support.google.com/analytics/bin/answer.py?hl=en&answer=1008080 ;-)

Suzanne
Re: Bounce Rate Attack fedesasso 1/10/13 7:04 AM
Klever Freire,

somehow your pages are incorporating both the asynchronous GA snippet, and the old synchronous one.
The second/synchronous one should be executed first, while the other is executed after the page is loaded as it is designed not to block rendering.

Execution of two GS calls would cause an incorrect zero bounce rate reading, as the article pointed out by Suzanneh explains.
Your analysis should I'm afraid be repeated once you gather enough BR history clean of the error.

Hope this helps.
Re: Bounce Rate Attack Lysis 1/10/13 7:07 AM
Dude you need to listen.... 1-3% bounce rate is not accurate at all. For instance, a 30% bounce rate is pretty good. 1-3% is a good conversion rate in some marketing campaigns. Maybe that's what you're looking at.
Re: Bounce Rate Attack smallbiz-Bob 1/10/13 7:13 AM
If you really want to know what is causing a high bounce rate, take a look at your Home page.

Your image slider is the problem.

First, the slider takes way too long to load. It takes 33 seconds to load. Visitors don't have that kind of patience and if you make them wait to see a picture show that they could care less about, they will bounce.

http://www.webpagetest.org/result/130110_N1_CQB/

Second, the slides change too rapidly and at an erratic rate. The first slide changes before you can read all of the text on it. The second slide changes to the third twice as fast. The timing of the others is erratic and WAY too short.

Your Home page also follows the "kitchen sink" approach. It is over-stuffed. Nobody is going to read through all of it. Less is more. Consider getting rid of the useless image fader entirely which eats up all of the valuable space above the fold, and put effective text in its place.
Re: Bounce Rate Attack Suzanneh 1/10/13 7:18 AM
Actually, you didn't copy all of the second instance.  This is what appears to be in your footer (or innerbody) and what needs to be removed:

--------------------
<script type="text/javascript">
var gaJsHost = (("https:" == document.location.protocol) ? "https://ssl." : "http://www.");
document.write(unescape("%3Cscript src='" + gaJsHost + "google-analytics.com/ga.js' type='text/javascript'%3E%3C/script%3E"));
</script>
<script type="text/javascript"> try { var pageTracker = _gat._getTracker("UA-18910772-2"); pageTracker._trackPageview(); } catch(err) {}</script>

-------------------

Suzanne


Re: Bounce Rate Attack Suzanneh 1/10/13 7:20 AM
>>For instance, a 30% bounce rate is pretty good.

I'd take that bounce rate any day and twice on Sundays!

Suzanne
Re: Bounce Rate Attack Klever Freire 1/10/13 9:09 AM

@Suzeanneh

I agree, it seems like the dual implementation of the track page view is odd.  Forgive me, but your explanation for how the code snippet is used by Google is not consistent algorithmically- but at first glance it does look like there is a possibility that a single page view would cause two page views to be tracked ON THE SAME PAGE for every legitimate bounce. If you believe BR = Tv/Te where Tv = total visitors (not unique visitors) and Te = total # of visitors viewing only one page...think about it algorithmically....

  1. page loads
  2. first instance of script runs tracking a page view but no exit is recorded (Tv = 1, Te = 0)
  3. second instance of script runs (Tv = 2, Te might then equal 1 or 0)
    • bounce may then be legit recorded

So, if Wordpress made an error in implementing the analytics code twice, then I will be recording 2 page views (of the same page) for every legitimate bounce. Bounce rate could not be 0, and mine never has been.  Aside from that, this does not explain how some days bounce rate jumps from 1%-3% to more than 44%. Your theory, if its correct does not explain how that can happen.  I believe, that no-one that has answered this post has explained algorithmically (with some source or proof) how google treats this apparently redundant instance of analytics page tracking code.  I may be wrong in my understanding, but theres no real proof besides quoting of articles that anyone else is correct either. This "long click attack" is inconsistent with the rest of my traffic- its an anomaly and shortly after bounce rate increases page rank is decreased by 5-10 spots.  Once my bounce rate recovers (to less than 10%) page rank is restored to top 3. 

I have no plug-ins that conflict with analytics page tracking.  If its a wordpress issue, then its an error made by the template provider that I'll have to investigate and correct manually.

Thanks everyone for your information- regardless of whether its constructive or not. 

Re: Bounce Rate Attack Suzanneh 1/10/13 9:42 AM
Remove the extra instance from all pages, and then come back to post your new bounce rates.  Prove me wrong. :-)

Suzanne
Re: Bounce Rate Attack fedesasso 1/10/13 9:46 AM
Dear Klever Freire,

your description of how BR is computed is flawed. Please let me explain better:

BR = Te/Tv (the inverse of what you wrote)
"Number of Visits with only one page-per-visit divided Total number of Visits"

Notice I'm talking about Visits, not page Views (nor visitors).
A Bounce happens when only one view is recorded per visit (session cookies are used to track multiple views within the same visit/session)

So if the GA call is performed twice, BR will be theorically zero:

  1. page loads
  2. first instance of script runs tracking a page view but no exit is recorded (Tv = 1, Te = 1)
  3. second instance of script runs (Tv = 1, Te = 0)
  • bounce is then computed as 0/1=0

The fact bounce rate has never been computed as an exact zero, but a to close number (compared to a normal BR value) can be explained with various hypothesis: the double GA code not being present in all of your pages, http calls gone lost, sessions expiring, etc...

Hope this helps
Re: Bounce Rate Attack Klever Freire 1/10/13 10:59 AM
@ fedesasso, all references I find for bounce rate show BR = Tv/Te. 

i.e. http://blog.kissmetrics.com/bounce-rate/
      http://en.wikipedia.org/wiki/Bounce_rate

I had a google issued reference as well, but it eludes my searching for now.

Re: Bounce Rate Attack Klever Freire 1/10/13 2:49 PM

Take a look at the suspected attack day metrics- January 4th and 6th.


Re: Bounce Rate Attack fedesasso 1/10/13 4:14 PM
dear Klever Freire,

> all references I find for bounce rate show BR = Tv/Te. 

That's because I adopted your notation where you defined:
> where Tv = total visitors (not unique visitors) and Te = total # of visitors viewing only one page

but if you look at the wikipedia page, the definition is invertedTv = Total number of visitors viewing one page only, Te = Total entries to page
Adopting wikipedia definition of Tv,Te, then Rb=Tv/Te but the result doesn't change:
  1. page loads
  2. first instance of script runs tracking a page view but no exit is recorded (Te = 1, Tv1)
  3. second instance of script runs, visits are still one because the two calls have the same session cookie but it's not anymore a single-pageview visit (Te = 1, Tv = 0)
  • bounce is then computed as Tv/Te=0/1=0
> I had a google issued reference as well, but it eludes my searching for now.
Also dig in the free resources of the Google Analytics IQ certification for an better understanding of GA metrics (visits, unique visits, visitors, page views, etc...) and how they are measured, and how derived metrics such as bounce rate and exit rate are computed.

I'm not saying you are wrong about the alleged attack, but you need correct data to be able to assess a situation; the sooner you fix your GA code, the sooner you be able to have it.

Kind regards
Re: Bounce Rate Attack Emma North 1/11/13 5:56 AM
Just to stick my two cents in here; the duplicate instances of Analytics tracking code is definitely the problem. A client of mine recently exhibited the exact same problem, highlighted when I noticed a bounce rate of 2.5% - something which is completely unnatural and obviously inaccurate.

The site had one correct version of the code and one amended version they were trying to establish e-commerce tracking with. The bounce rate becomes unpredictable, with a consistently low (under 3%) bounce rate for nearly two years before some random jumps and drops early this year.

The bottom line is, if you wish to see accurate bounce rates, you must get your tracking code right.

And just to clarify, bounce rates recorded in Analytics will not affect your rankings in any way. Google does not use Analytics tracking code or recorded data in it's algorithm.

Hope this helps :-)
Re: Bounce Rate Attack robdwoods 1/11/13 3:22 PM
Seems to be an oddly low bounce rate number too... 1% bounce rate? I don't know any site that has a 1% bounce rate. That seems crazy low. Maybe I'm wrong...

I recently had a site where I had the analytics code duplicating pageviews causing analytics to report a 6% bounce rate and when we fixed the issue the bounce rate when to a more realistic 30%.

I wonder if this is an "attack" which would have no point anyway, or if they actually made a change to their site or analytics that is reporting a more accurate bounce rate now.
Re: Bounce Rate Attack WillieO 1/11/13 10:27 PM
It has no impact whatsoever as far as Googles concerned. Some of the best performing ( 18% conversion rate) websites I've done have a bounce rate of over 50%. Why would that be considered good? Because I practice specificity on a per page level. When a user does a search they land on that specific page and they call. A phone call being one of the conversions, the other being landing on the contact page. There's no need for them to proceed to other areas of the website. They found what they were looking for.
 Alot of people smarter than I have given you good advice here. 
Susan hit the nail on the head with the double analytics code.
Beusserys advice should be taken.

Re: Bounce Rate Attack Suzanneh 1/12/13 4:26 AM
I don't sell a product; my site's revenue is ad driven.  One of my best performing pages, organic search traffic-wise, has a 78% bounce rate average.  That's because it's a particular calculator that some people use every day.  They come in, get the info and go on with their lives.  It's been in a #1 position for several different keyword phrases for as long as it's been put up.

Suzanne
More topics »