Sorry this site requires JavaScript to be enabled in your browser. See the following guide on How to enable JavaScript in Internet Explorer, Netscape, Firefox and Safari. Alternatively you may be blocking JavaScript with an advert-related or developer plugin. Please check your browser plugins.

The relationship between bounce rates and SEO have been discussed at Sphinn and elsewhere lately. This post brings together 9 major objections to linking SEO and bounce rates, and refutes 8 of those 9 objections.
Comments31 Comments  

Comments

Avatar
from bwelford 2036 Days ago #
Votes: 1

Clearly there is information content in how visitors navigate around websites.  Google and the other search engines can easily capture a good part of that information content.  As usual, whether and how they do it they will keep under wraps.  In the same way we all consider carefully how we package the content on the web page, we should also consider how the bouncing phenomenon, however defined, may influence rankings.  I hope that’s clear. :)

Avatar
from theGypsy 2036 Days ago #
Votes: 6

<!--[if gte mso 9]><xml> <w:WordDocument> <w:View>Normal</w:View> <w:Zoom>0</w:Zoom> <w:Compatibility> <w:BreakWrappedTables/> <w:SnapToGridInCell/> <w:WrapTextWithPunct/> <w:UseAsianBreakRules/> </w:Compatibility> <w:BrowserLevel>MicrosoftInternetExplorer4</w:BrowserLevel> </w:WordDocument> </xml><![endif]--> <!-- /* Font Definitions */ @font-face {font-family:SimSun; panose-1:2 1 6 0 3 1 1 1 1 1; mso-font-alt:宋体; mso-font-charset:134; mso-generic-font-family:auto; mso-font-pitch:variable; mso-font-signature:3 680460288 22 0 262145 0;} @font-face {font-family:"@SimSun"; panose-1:2 1 6 0 3 1 1 1 1 1; mso-font-charset:134; mso-generic-font-family:auto; mso-font-pitch:variable; mso-font-signature:3 680460288 22 0 262145 0;} /* Style Definitions */ p.MsoNormal, li.MsoNormal, div.MsoNormal {mso-style-parent:""; margin:0cm; margin-bottom:.0001pt; mso-pagination:none; mso-layout-grid-align:none; punctuation-wrap:simple; text-autospace:none; font-size:10.0pt; font-family:"Times New Roman"; mso-fareast-font-family:SimSun; mso-font-kerning:14.0pt; mso-fareast-language:ZH-CN;} @page Section1 {size:612.0pt 792.0pt; margin:72.0pt 90.0pt 72.0pt 90.0pt; mso-header-margin:36.0pt; mso-footer-margin:36.0pt; mso-paper-source:0;} div.Section1 {page:Section1;} --> <!--[if gte mso 10]> <style> /* Style Definitions */ table.MsoNormalTable {mso-style-name:"Table Normal"; mso-tstyle-rowband-size:0; mso-tstyle-colband-size:0; mso-style-noshow:yes; mso-style-parent:""; mso-padding-alt:0cm 5.4pt 0cm 5.4pt; mso-para-margin:0cm; mso-para-margin-bottom:.0001pt; mso-pagination:widow-orphan; font-size:10.0pt; font-family:"Times New Roman"; mso-fareast-font-family:SimSun;} </style> <![endif]-->  Seriously, enough already. I appreciate the effort but this does nothing to discount any realities surrounding bounce rates as a serious signal.   It’s funny, guys like Bill (and yer truly) have been writing/talking about behavioural metrics in search for more than a year and just now the community latches onto it and starts professing their find. After reading dozens of papers and patents (not to mention TESTING) the realities are simply obvious to me.   To start with, bounce rate is but one of many behavioural metrics search engines look at (both implicit and explicit signals). Why is it that peeps that read a few posts on bounce rates seem to be fixated and look no deeper? Laziness? Just looking to make headlines? Or worse? Why no talk of the other signals?   Until peeps writing about it have done the due-diligence in researching ALL the implications of behavioural metrics as ranking signals, I feel they shouldn’t be making statements such as;   “I am of the opposite view; bounce rates will certainly be a major part of search engine algorithms and probably already are to some degree. ”   Amazing… I just don’t get it? There is NO WAY it will play a ‘major’ role as a ranking signal in a standalone capacity (void of other behavioural signals) within the regular index (non-personalized environment).   This post is simply not well researched and smacks of attention getting, not a true, well thought out, post on indexing and retrieval. SEOs simply MUST start to learn more about IR, probabilistic modelling, NLP and more. This would bring a more balanced view to our subjective assertions.

Avatar
from amabaie 2036 Days ago #
Votes: 1

Hi theGypsy.  You are not taking into account the full story.  There is no suggestion that bounce rates is a stand alone metric, any more than PageRank is.  The most important aspect of my article is in fact how the search engines can read a user’s activity after having recorded a bounce.  In fact, that is crucial.  It is only once someone has been to a site that the search engine can get feedback from the person.  That feedback can come in many forms.  A bounce is the simplest feedback.  The lag time of the bounce is probably the next simplest.  The next action is more complex (click on another result of the old search, new search almost identical, new search somewhat similar, new search completely different, no new search, new search engine - yikes!, close window, etc.)  Harder to measure (from a privacy perspective) but also great data would be how the user moves around the website (fills form, visits multiple pages, closes window, moves to am external link, etc.)    All these are great signals, and as I said, a bounce does not stand on its own as a metric.  That being said, no signal is as strong a measure of customer satisfaction and as easy for a webmaster to measure as the simple bounce.  I find that it is almost always helpful when analyzing things online, to find a parallel in the real world.  A bounce is someone leaving the store.  If a person leaves the store after just a few moments, clearly the store did not satisfy them.  If most people leave the store without buying, the mall owner might want to move the store to a hidden corner and put a higher-satisfaction store in its place so people keep finding the mall useful and keep shopping there.The only signal more potent than leaving the store after a few seconds would be leaving the store after a few seconds while sticking out one’s tongue and saying  “brzktftx!”.  Maybe Google can create an online tongue-sticking tool to help make it even more clear which links to relegate to the corners of its "mall".

Avatar
from theGypsy 2036 Days ago #
Votes: 5

Ok let’s play this game then... Where is the data on collection points? What percentage of sites use Google Analytics? How many ToolBars are there? The simplest problem is that using such a metric as a serious signal means there has to be large enough data sets - which there isn’tHow about related metrics? What of explicit and implicit signals? The post seems like a noob offering that does liitle to dig into how behavioral metrics work and I suspect they don’t fully understand them. That being the case how can such statements be made? It is simplistic at best.And what of Spam? I see nothing there that is actionable by Matt and company. Do you really think it is a simple as ’hey let’s use this signal’ - bouce rates and behavioral in general make for noisy signals that are problematic from a spam management perspective. And what of bookmarks? Application focus? These are tools that ’could’ (in concert with other approaches and signals) make behavioral more viable, but even those methods are still a work in progress and this all bloats the processing requirements and ultimately cost. I’ve been reading, researching, studying and testing behavioral metrics for more than a year - and there are far greater minds than mine working in this area to mine the best signals and situations for these metrics to be used in search quality. This post addresses none of the real-world problems associated with behavioral metrics outside of a personalized setting.This is how mythologies start in the world of SEO - biased and unresearched supositions that lean more towards assertions than hypothesis from SEOs. I had a client go to PubCon recently and spent 3 days de-programming him from the mythologies he picked up from SEOs that are either ignorant or egotistical over methodical. This kind of stuff DOES cost me time/money.So they can go back to the drawing board and actually learn about IR issues surrounding explicit and implicit user signals... just sayin’ (and I love SEO... so I ain’t dissing here... just love my search)Seriously, I once wrote about the future of behavioral metrics in search... over the last year my views have tempered as I learned more about the systems and problems they face.All I ask is that peeps stop making assertions about things just because of their ’gut feelings’ - a ’major’ or ’serious’ signal it is not... Jeez, even Google said so.. and don’t give me the "thats what they would say’ as when it suits us we believe their word to be gold... the SEO FUD starts when peeps think their suspositions will somehow position them as SEO Experts...

Avatar
from theGypsy 2036 Days ago #
Votes: 2

So just for the sake of it... peeps might want to read; http://www.huomah.com/Search-Engines/Algorithm-Matters/Is-Google-REALLY-using-bounce-rates-as-a-ranking-signal-.htmlAnd certainly the 10 other posts linked at the end of it relating to behavioral metrics in modern search. /nuff said... I am a huge fan of behavioral metrics and would be happy to chat about it or pass along some resources to anyone interested. amabie that includes you. I ain’t lookin to be a dick,  just worry about assertions as I love my SEO as well...

Avatar
from amabaie 2036 Days ago #
Votes: 0

The data collection point is every Google searcher and every action they take.  Every time they leave Google to follow a link and every time they return to Google.  It is every time they make a new search, click to a second page of results or perhaps even scroll down the page.  Those are the data points.  The toolbar and Analytics and bookmarks and other tools for measureing user actions might give some supplemental data, certainly helping with Google set aggregate benchmarks and perhaps helping them verify their assumptions about their algorithm, but discussions of these tools is a distraction from what really counts - the customers (that is you and me when we are searching, not when we are webmasters). I think you are over-thinking this, trying to find technological problems.  The bottom line is that the search engines are all competing for customers.  Like any sector, the ones who best satisfy the customers win.  It’s not a matter of will they or won’t they measure and account for user metrics.  In my view, that discussion is not even on the table.  There is nothing egotistical about such an assertion; it’s business 101. 

Avatar
from iBrian 2036 Days ago #
Votes: 0

David, the problem is that you are simply making personal objections based on personal perception of the issue, rather than the real business 101 of looking for signals in the wild.There’s the assertion "However, none of the tests and reports seem to be complete enough or repeatable enough to constitute “proof"However, this pretty much sums up most of SEO.A lot of people have blogged about seeing potential instances of bounce rates being used as a potential signal, and all of us can be held to be unable to prove it, but we share what we think we are seeing for general consumption.No one has to accept those observations, but to dismiss out of hand due to personal aesthetics rather than signals from data probably isn’t the wisest policy in SEO.

Avatar
from theGypsy 2036 Days ago #
Votes: 2

Personal objections on personal supositions by others. My personal issues are related to a vast amount of research as well as what we do know from the SEs. I am just amazed as time passes how SEOs seem to care little for the realities of search evolution over what is ’sexy’ - once more, I was singing the bahavioral metric songs long ago... so it’s funny that something I once touted I am now speaking against, (in that the current theories are limited views). Once more, it’s not that I am poo poo’ing the concepts, pretty sure you’ve read me talking about this many times over the last year.. problem is that I have had more time to consider elements and potential limitations. How does one do that? Not reading Hot stories on Sphinn I can tell U that much. It’s about learning and testing as you know... and hyperbole doesn’t compare to some good ol fashion IR studies to broaden ones knowledge in understanding what they’re looking at (reverse engineering).I’ve said it a few times lately... I am saddened by the lack of technical IR knowledge of many SEOs... We really are link baiting hype merchants more than technicians. Each has it’s place certainly, but we are sadly lacking on the technical training. I couldn’t imagine a web developer that doesn’t know HTML but SEOs that understand search egineering, even in passing, is troubling.I think that’s what’s eating me bri, I could really care less what other SEOs believe for the most part, if I’m right then my time (and client’s money) are being invested wisely and I shall continue to dominate query spaces. This is near business intelligence, so fine by me.  Just as a passionate fella about the ’Art’ I tend to get frustrated not having actual educated discussions of IR technologies. I mean.... without searching, who can tell me the various behavioral signals search engines might look at? What are the data collection points? And more;A. How do you (the search engine) know that an apparent bounce is a negative? What if I bookmarked it? Gave to a friend in chat or marked in a Word Doc (application focus)? - What if I found what I wanted on the first page? Whereas another satisfied searcher went to other pages? Proving satisfaction with mere bounce rates is not strong indicator without other signals in place. B. Data collection beyond mere ’backpage’ is not really going to be effective. Google has talked about collection points such as toolbar data, email, IM (application focus) and others. There are though deployment and saturation issues that still don’t lend themselves to a truly global data set that could be used as a strong signal.And that’s really the tip of the iceburg... we haven’t even gotten into the spam related issues.Once more, I merely object to inferences of bounce rate being a ’serious’ or ’major’ signal. I have no doubts of it’s use, moreso in a personalized setting, I am merely disputing the assertion it is that valuable a stand alone metric within the regular SERPs... there are are far better ones (behavioral metrics) TBH... just no one learns about it... no one talks abou it... so SEOs play with what they have available... what is so wrong with me saying ’learn the rest of the story - then make your hypothesis’?As for ’Business 101’ (amabaie) - great... but I am more interested in talking IR 101 :0) - "I think you are over-thinking this, trying to find technological problems.  The bottom line is that the search engines are all competing for customers. "Yes, and believe it or not, much of that is getting search quality RIGHT. Which means they are always looking for techincal problems. That is what search engineers get paid for, not spreading FUD as our narcissistic industry likes to believe. I agree behavioral metrics have value and are being used in search today, I am merely voicing cncern over jumping the gun and asking peeps to learn more about the topic and wide variety of signals that make up this area... Anyway, enough of this... I am sure there will be plenty of this to yak about in the new year... only took a year...lol... glad we’re at least talking about it.Peace ya’ll... time for some holiday cheer - Happy holidays and thanks for the solid chat.. Healthy stuff...

Avatar
from iBrian 2036 Days ago #
Votes: 0

Apologies, I’ve just realised both of you are called David - my earlier comment was directed to David (amabaie) - I’m in full agreement with David (theGypsy).Just thought I’d clarify. :)

Avatar
from theGypsy 2036 Days ago #
Votes: 0

NP Bri...tnx for clearing that up... Ho ho ho 2 U and yours my friend.

Avatar
from amabaie 2036 Days ago #
Votes: -2

OK, I quote from the other David.  :-)   "How do you (the search engine) know that an apparent bounce is a negative? What if I bookmarked it? Gave to a friend in chat or marked in a Word Doc (application focus)? - What if I found what I wanted on the first page? Whereas another satisfied searcher went to other pages?"  This is great.  These are "how" questions.  These are the questions that a search engine executive would ask his engineers.  In responding to these questions, one creates an algorithm.  If you look at the average time people spend on a page they send to a friend or bookmark a page or do anything with, it should be longer than if somebody looks around, doesn’t see what he wants and hits the "back" button.  Yes, it is possible that a few super speedy people land on a page, instantly make a judgment without even reading it and forward it to a friend, then bounce back.  These are statistical aberrations and there are tools to control for these.  Typically in any survey of reasonable proportions, one removes the outlier stats (I’ve usually seen the highest and lowest 5% of responses removed).  That way statistics are not skewed by a few people who act with lightening quick speed or a few who leave the window open while eating dinner and then bounce back when they return, for example. On average a person reading a page will stay there longer than a person who chooses not to.  On average a person doing something with a page will stay there longer, having most likely read the page, than someone who just reads.  Not always, but that doesn’t matter because... Remember, we are dealing with aggregate numbers.  The algorithm does not have to accurately determine your intention on any given occasion.  It has to be right often enough to form an accurate overall picture.  With a search such as "speckled peanut tanning lotions" they might not have enough search volume to be able to apply user metrics with an acceptably high confidence level.  Again, that is part of the "how".  The search engineers are responsible for determining how best to account for exceptions and make sure the algorithms work within whatever tolerance levels they set for themselves.   Just because any of us can come up with aberrations, does not mean something can’t be done.  In fact, I could come up with as many objections to suggest that PageRank is impossible...or any other part of the current algorithms. I think we agree that they have to get the technical aspects right, and I am sure you know much, much more about that than I do...but again, that is the "how".  Technology is an enabler.  It is about the "how".  The goal is to deliver results that satisfy the customer enough to return to the search engine rather than switch to a competitor.  I maintain that bounce rates are one measurement among several, or part of the measurement, that cannot be ignored.  That’s all; nothing more nothing less.  So...we’ll continue this in the New Year, I presume.  :-)

Avatar
from Feydakin 2036 Days ago #
Votes: 1

Bounce rate can be so skewed that it makes for a terrible ranking metric.. A prime example is a friend’s website.. http://www.google.com/search?q=how+big+is+a+millimeter&sourceid=navclient-ff&ie=UTF-8&rls=GGGL,GGGL:2006-34,GGGL:en&aq=tHe ranks #1 for that search because it is important for the product he sells.. But, the people that come to his site to find out what a mm is hit the page, and leave in less than a minute 95% of the time.. They got exactly what they wanted but it crushes his bounce rate.. So how could that be a good signal for ranking in this scenario?? Could a high overall bounce rate for his site, because of this one page, affect his overall rankings?? I hope not, but we plan to do some testing after the first..

Avatar
from MattCutts 2036 Days ago #
Votes: 3

Without reading the article, I’ll just say that bounce rates would be not only spammable but noisy. A search industry person recently sent me some questions about how bounce rate is done at Google and I was like "Dude, I have no idea about any things like bounce rate. Why don’t you talk to this nice Google Analytics evangelist who knows about things like bounce rate?" I just don’t even run into people talking about this in my day-to-day life.

Avatar
from robwatts 2036 Days ago #
Votes: 0

Matt, that’s an interesting statement ;)So bounce rates are a no factor, but what about all the other onpage user engagement actions?If you are suggesting that smart people wouldn’t try to fold in aspects of user behaviour and engagement into a system so otherwise susceptible to elements that are highly manipulable then that would be pretty huge.But I guess, you aren’t saying that, or are you?

Avatar
from amabaie 2035 Days ago #
Votes: 0

Matt, do you have any comment on Microsoft’s BrowseRank paper?  Or is that too loaded a question? 

Avatar
from tonyp 2035 Days ago #
Votes: 0

Matt thanks for the clarification!  Sometimes we over analyze the subject.

Avatar
from SeriesOfTubes 2034 Days ago #
Votes: 0

To use Bounce rate as a ranking factor, is extremely close to using adsense CTR and eCPM to determine organic SERP .   Google can’t really go there, because if they started ranking based upon these factors, the Govt. would raise the monopoly issue.That larger hurdle is ’user intent".   Until they nail user intent, they will not play with their search product by data sampling.I see this as an industry issue that SEO’s use to extract further money from clients, or more likely use it as scare tactic to steal business from competitors.   What can they say now?  I’ll up your action rate .02?  That won’t work so they convince the prospect that with a 77% bounce rate Google’s going to deindex you.   Merry Christmas.

Avatar
from mileusna 2034 Days ago #
Votes: 0

I use toolbar to search Google, so my Google entry page is SERP page. I quickly pick up the result and navigate to some other page away from Google.So my bounce rate on Google.com is around 90% I guess. Does that mean that Google.com is lousy website since the bounce rate is so high? :)

Avatar
from iBrian 2033 Days ago #
Votes: 0

"bounce rates would be not only spammable but noisy"Isn’t this all search signals, until some form of filtering is applied? :)

Avatar Moderator
from incrediblehelp 2033 Days ago #
Votes: 1

As in the Feydakin example above using BR as a ranking factor doesnt make much sense.  Many websites out there have some terrible BR rates for certain keywords.  Does that mean the website is any less usable or justified in achieving ranking?  Not at all in IMHO.  Some websites have ads on their pages right?  Getting the user to leave through those ads is a success for the website owner and might be the primary business model for the website.

Avatar
from seobro 2032 Days ago #
Votes: -1

I have already started to make changes to keep bounce rate low. You can add new stories a lot and place a large red NEW button in front of them. Also sensational stories to make people click like "Shark devours woman". "Vegetarian diet lowers IQ", "Boy claims to have been Julius Caesar in previous life".

Avatar Moderator
from Jill 2032 Days ago #
Votes: 2

Guys, does it really matter whether Google uses bounce rate as a signal? If you have an overly high bounce rate there could be something wrong with your site (or certain pages of it). That’s what’s important, no? Stop worrying about what Google thinks and use all the information at your disposal to fix your site! Doing that will magically make Google like it better too. As to this thread specifically, I very much liked and mostly agreed with TheGypsy’s posts. Especially like the reprogramming of the client after going to PubCon. So true, so true!

Avatar
from g1smd 2032 Days ago #
Votes: 0

Of course it matters. Were it to be true, there would be a whole load of people looking to fake the figures for their sites, and by tomorrow a whole new industry would have sprung up offering fake visits to your website, and guaranteeing to visit multiple pages, to "enhance your SEO".  :barf: 

Avatar
from theGypsy 2032 Days ago #
Votes: 0

he he... @Jill... sad task ain’t it? That’s the only reason I go off as well as G’s mention of the wacky services that pop up. Ultimately when I went sideways on this a few weeks back it was exactly that, watching the birth of (yet another) mythology. And hey... I did mention that it was more important as a matter of good webmastering not new groovy signals to purvey SEO fud :0)@G ... hopefully it won’t come to that, but undoubtedly there will be other equally ignorant services popping up anyways..lol...

Avatar Moderator
from Jill 2032 Days ago #
Votes: 0

@g1smd eeek I hadn’t even thought of that. Ugghh...they’re probably already out there, come to think of it.@theGypsy well you know that my pet peeve is SEO myths so I’m very happy to bust as many as possible and it’s always good to see others doing the same!

Avatar
from EiSa 2030 Days ago #
Votes: 2

From an interview with Risvik, Head of Google Engineering in Norway:It can also be taken account for how long you visit a site. According to Risvik they can measure how many seconds it goes from a user clicks on a link to that person clicks back to Google.The whole article (translated by Google):http://translate.google.com/translate?hl=en&u=http%3A%2F%2Fwww.digi.no%2Fphp%2Fart.php%3Fid%3D798287&sl=no&tl=en

Avatar
from donovanroddy 2030 Days ago #
Votes: 0

WOW - good discussion, theGypsy obviously had a six shot Mocha before posting here!

Avatar
from wredlich 2023 Days ago #
Votes: 0

Some of my pages have high bounce rates but do well in searches. I think "returning visitors" might be a more important metric, if search engines are going to use such metrics. More than half of those who hire me (my law firm website is the source of our business) are returning visitors.

Avatar
from philippe 2016 Days ago #
Votes: 0

I disagree with the original definition of a bounce rate in  your post "when people click the back button". The bounce rate is the share of single page views visits divided by total number of visits. The back button is only one share of it (people might have gone to a bookmark, closed the browse, etc. Also, unfortunately we can not tell how long a visitor stayed on your website after a single page visit: therefore we can not talk about a 10 seconds bounce rate, or a 25 minutes bounce rate, as simple as we CANNOT know this information. It does not make any sense to talk that way.

Avatar
from corey 1835 Days ago #
Votes: 0

original article takes forever to load waiting for spicypage.com i’m sick of waiting.

Avatar
from philippe 1835 Days ago #
Votes: 0

The post is based on assumptions, and simple definitions are actually wrong.

Upcoming Conferences

Search Marketing ExpoSearch Engine Land produces SMX, the Search Marketing Expo conference series. SMX events deliver the most comprehensive educational and networking experiences - whether you're just starting in search marketing or you're a seasoned expert.



Join us at an upcoming SMX event: