SEO Blog - Internet marketing news and views  

The final word on bounce rates as a ranking signal

Written by David Harry   
Thursday, 15 January 2009 08:22

Putting behavioural metrics in perspective

Wise men don't judge: they seek to understand. - (Wei Wu Wei)

So here’s the question; are behavioural metrics being used in modern search? You do remember them right? Those warm and fuzzy little signals such as bounce rates that there all the rage in late 2008 in the search engine optimization world? Sure you do… but let’s take one last look.

Although bounce rates received the biggest attention, we would be remiss not to start by quickly listing some signals commonly looked at by information retrieval folks. The two elements include implicit and explicit data (actions and interactions) – examples can include;

Implicit signals

  1. Query history (search history)
  2. SERP interaction (revisions, selections and bounce rates)
  3. User document behaviour (time on page/site, scrolling behaviour);
  4. Surfing habits (frequency and time of day)
  5. Interactions with advertising
  6. Demographic and geographic
  7. Data from different application (application focus – IM, email, reader);
  8. and closing a window.

Explicit signals

  1. Adding to favourites
  2. Voting (a la Search Wiki or toolbar)
  3. Printing of page
  4. Emailing a page to a friend (from site)

Now that we’re past that let’s get a little geeky so those information retrievers don’t shake their heads to hard at us – the terminology. I am as guilty as the next Gypsy of flinging the term ‘behavioural metrics’ about over the last year or so, even performance metrics. If you want to research this more, start by using the term; implicit/explicit user feedback signals – because that’s what we’re talking about.

This is not the ranking signal U were looking for
and thanks to Steve Gerencser for sending the pic

Follow the bouncing timeline

While you can trace the timeline back in the search (blogging/reporting) world many years, it really came home when Search Engine Watch mentioned it (Oct.2008) followed a few months later in a Search Engine Land post. Being the venerable publications that they are, grumblings around the SEO world soon followed. If you go and do some buzz monitoring and searching (which I have), much of the talk began after that. The cracks in the damn began to fissure and this Gypsy was left without enough chewing gum.

So what can we do? Where does one start to truly look for answers as to the potential of such methods being implemented by top public access search engines? It would stand to reason that we begin looking at the information retrieval world itself. Over the last month I have given the benefit of the doubt to the community and gone deeper to find some type of more definitive answer, (list of research papers at the end).

 

Inherent problems with implicit signals

One thing that became obvious real fast is that the IR world is still not entirely sure of the value for implicit feedback signals as far as how to infer engagement and satisfaction. While there are a long list of problematic areas let’s consider;

  1. You save the link for later and continue my search (in Doc let's say)
  2. You found what u needed on the page and went looking for more information
  3. You walk away from my browser and leave the window on a page for an hour
  4. Multiple users in your home during a given session
  5. Open a listing in a new window (when further tracking is unavailable)
  6. You found the information in a SERP snippet and selected nothing
  7. You were unsatisfied with the page selected and dug 3 pages deeper (unsatisfied, not engaged)
  8. Queries from automated tools (like a rank checker) which adds noise to overall data
  9. SERP bias – do peeps simply click the top x results regardless of relevance?
  10. Different users having different understanding of the relevance of a document (result)

…and on and on. Think about it, some situations can tell the search engine you’re pleased with the results and other times such signals mean nothing. You see, the essential motive is to attempt to assign an emotional evaluation of engagement with the search results. Unfortunately there are too many noisy elements which make this a very difficult task to do effectively.

Noise and confused

It’s widely felt that ‘implicit feedback is more difficult to interpret and potentially noisy’ as noted in - Evaluating the Accuracy of Implicit Feedback from Clicks and Query Reformulations in Web Search (partially funded via grant from Google) – in looking at click behaviour there was indeed a clicking bias based on a few elements;

“….First, we show that there is a “trust bias” which leads to more clicks on links ranked highly by Google, even if those abstracts are less relevant than other abstracts the user viewed.

Second, there is a “quality-of-context bias”: the users’ clicking decision is not only influenced by the relevance of the clicked link, but also by the overall quality of the
other abstracts in the ranking.”

Other research (on click data) looked at how users actually interact with search results as far as bias is concerned. People are often consistent in clicking patterns (clicking top result, second, third) regardless of the underlying data. This means the entire data set can be skewed as not clicking on the 8th result may no necessarily be a vote against the link in the result, but more of an ingrained habit on the part of the searcher.

They summarized;

“Our results show that click behaviour does not vary systematically with the quality of search results. However, click behaviour does vary significantly between individual users, and between search topics. This suggests that using direct click behaviour—click rank and click frequency—to infer the quality of the underlying search system is problematic.”

And also;

“Analysis of our user click data further showed that the action of clicking is not strongly correlated with relevance —only 52% of clicks in a search result list led to a document that the user actually found to be relevant. Attempts to use clicks as an implicit indication of relevance should therefore be treated with caution.” From - Using Clicks as Implicit Judgments: Expectations Versus Observations

Beyond that many of the papers have various elements of implicit user feedback that they felt warranted more study. In short, there is no consensus in the IR community about the validity of these signals – they’re not ready for prime time.

ok. 

The Spam connection

And this my friends, as they say, is the proverbial fly in the ointment. While there is a ton of research and even patents on behavioural metrics, dealing with click-spam has not been addressed in any detail to this point. Many of the papers openly admit they are light in the spam detection area and more research is needed.

 
A natural question that arises in this setting is the tolerance of this method to noise in the training data, particularly should users click in malicious ways. While we used noisy real-world data, we plan to explicitly study the effect of noise, words with two meanings, and click-spam on ourapproach. From - Query Chains: Learning to Rank from Implicit Feedback

And that’s just one; it was a common theme among the papers on the topic. This for me goes a long way into understanding that it is premature to suggest search engines that we optimize for are using such signals. There is hope as some tests, as ran by Microsoft, concluded;

“ranking accuracy decreases indeed when more documents are spammed, but the decrease is within a small range. When only a small number of documents are spammed per query, ranking accuracy is only slightly affected even if a large number of queries are spammed.” From- Are click-through data adequate for learning web search rankings?

They felt that such a large percentage of queries are long tail queries that it would be more difficult to effectively disrupt the majority of query spaces (I hear Ralph rumbling some where with that one). But once more, there seems to be a lot more work to be done in this area to effectively combat spam in such a system. To this we add thoughts from a Cornell paper;

“… it might also be possible to explore mechanisms that make the algorithm robust against “spamming”. It is currently not clear in how far a single user could maliciously influence the ranking function by repeatedly clicking on particular links.” From - Optimizing Search Engines using Click through Data – Cornell (pdf)

For me, there simply isn’t enough research or hard data to suggest that the spam issues related to implicit user feedback and click data have been solved. This is a crucial element to the case of them being used today by Google or anyone else.

Not enough, then also try this recent post by your friend and mine, CJ, on Clickstream spam detection or Fantomaster’s Behavioral Metrics and the Birth of SEO Surfbot Nets – let us get to then now shall we?

 

Getting beyond the geeky; looking to the future

Are we getting somewhere yet? Great… but it’s not all doom and gloom, no need to call the corner just yet. You see, for the most part researchers have been finding some great improvements in search performance; they simply haven’t worked out all the values of such signals nor the spam concerns. In an enterprise environment, where manipulation/spam is far less likely, implicit feedback can be a more useful tool. It is the larger public access environment where spam is far more prevalent that the nut has yet to be cracked.

I stand on my original assertion that this type of approach is best served in a personalized environment. This would be huge in dealing with the apparent issues surrounding spam related issues as it is kinda’ hard to spam ones self you see. This makes personalized a likely candidate for user feedback signals. Either way, it simply hasn’t been solved yet

So what are we left with?? Some noisy signals that are spammable… hmmm… where have we heard that before?

Matt Cutts on bounce rates

And so now I leave all of this in your capable hands my weary web warriors. If you can go through the research papers listed below (or elsewhere) and find me strong evidence of how they deal with noise reduction and click-spam, then we can discuss it further. That is my challenge to you; because from what is out there, it is not yet viable in a large scale environment.

I submit to you, my enthusiastic optimizers, that bounce rates and it’s implicit feedback brethren are simply not likely to be in Google’s (nor any major search engine's) current ranking schemes. It is a novelty item at best with potential in a personalized environment.

Care to dispute this? I am more than happy to review any research to the contrary.

Want to know what I think is causing us to see what we believe this to be? You’re just going to have to wait until next week.

 

/story..

“Muddy water, let stand becomes clear.” - Lao Tzu

 

Research looked at for this post;

Using Clicks as Implicit Judgments: Expectations Versus Observations - RMIT

Improving Web Search Ranking by Incorporating User Behaviour Information - Microsoft

Evaluating the Accuracy of Implicit Feedback from Clicks and Query Reformulations in Web Search - Cornell (funded in part via grant from Google)

Learning user interaction models for predicting web search result preferences - Microsoft

Improving rankings in small-scale web search using click-implied descriptions - ICT Centre

Identifying “Best Bet” Web Search Results by Mining Past User Behavior - Microsoft

Using Clickthrough Data to Improve Web Search Rankings - Algorithms for Data Base Systems Seminar,

Query Chains: Learning to Rank from Implicit Feedback - Cornell

Are click-through data adequate for learning web search rankings? - Nankai University,

Automated Evaluation of Search Engine Performance via Implicit User Feedback - Pennsylvania State University

A Comparison of Evaluation Measures Given How Users Perform on Search Tasks – RMIT University

Modelling A User Population for Designing Information Retrieval Metrics (dec 2008) - Microsoft

Bayesian Adaptive User Profiling with Explicit & Implicit Feedback - USC

Users’ Effectiveness and Satisfaction for Image Retrieval – University of Sheffield

Accurately Interpreting Clickthrough Data as Implicit Feedback - Cornell

Active Exploration for Learning Rankings from Clickthrough Data - Cornell

Web Search Engine Evaluation Using Clickthrough Data and a User Model - Yahoo

 

Spam Related;

Identifying Web Spam with User Behavior Analysis - Tsinghua University

Are click-through data adequate for learning web search rankings? - Microsoft

Optimizing Search Engines using Click through Data – Cornell (pdf)

Be sure to visit - Fourth International Workshop on Adversarial Information Retrieval on the Web (2008) and the 2007 stuff

 

Not enough for you? Here are some videos to pass the time away…

 

Videos;

Implicit feedback learning in semantic and collaborative information retrieval systems - Gérard Dupont, EADS, EADS

This presentation try to provide an overview of one way to resolve those gaps: using feedback learning. The aim is to make the system learning on user behaviour in order to better define its current needs. Machine learning algorithms applied on signal coming from user while performing a search can lead to the understanding of what is really relevant to the users and then can be exploited to help him during its tasks.

User models from implicit feedback for proactive information retrieval - Samuel Kaski, University of Helsinki
Our prototype application is information retrieval, where the feedback signal is measured from eye movements or user’s behavior. Relevance of a read text is extracted from the feedback signal with models learned from a collected data set. Since it is hard to define relevance in general, we have constructed an experimental setting where relevance is known a priori.

 

Proactive Information Retrieval by User Modeling from Eye Tracking - author: Jarkko Salojärvi, Helsinki University of Technology  

I now leave this to you.... where do you stand? Are search engines using behavioral metrics?

 

Comments  

 
-4 # Feydakin 2009-01-15 08:29
Ding dong the witch is dead! Death to bounce rates as a metric for ranking. Besides, I heard that bookmarking a page was the new metric that Google was using by tracking what bages are in your bookmarks.htm file.
Reply | Reply with quote | Quote
 
 
-2 # Richard Fergie 2009-01-15 08:36
I came across a blog post from a guy who teaches data mining at Stanford. anand.typepad.com/datawocky/2008/03/more-data-usual.html He explains how in his experience considering data of different types is better than getting more data of a single type. Google collect bounce rate data and have a history of using new data signals; I think they use it, but they might not assign it much importance.
Reply | Reply with quote | Quote
 
 
+1 # Dave 2009-01-15 09:09
OK Richard, one's feeling that more signals make better information retrieval have no bearing on this issue. Once more, search engineers have and are struggling with implicit feedback and some even believe it is dead horse.

You are making the first mistake in merely focussing on one element (bounce rates). The problem is assessing the intent and thus the value of such signals. While research has whown some promise, there are still problems.

So, please don't make the assertions with 'gut feelings' its a disservice to all in the SEO worled - do some homework first and bring something more to the table...

It simply is NOT going to be in the playbook yet....

just sayin'
Reply | Reply with quote | Quote
 
 
+2 # Richard Fergie 2009-01-15 09:54
I agree that recognising signals from implicit feedback is a hard problem (and the links you've provided indicate that a lot of experts think the same).

I also think that, if we didn't know about pagerank, we'd say that recognising links as a signal of relevance would be a hard problem as well. I still think it is, but the Search Engines seem to get by.

I'd also like to apologise for jumping into this discussion unprepared. I think I have a lot of reading to do
Reply | Reply with quote | Quote
 
 
0 # Carps 2009-01-15 09:30
..and on the other hand (and stepping away from the science for a second, compelling though it may be)...

Google bought Analytics software, and positively beg you to install it on your sites. That's hard usage data that goes far beyond mere bounce rate that Google can access for individual sites and aggregate across markets. Throw in goal tracking, event tracking, site search tracking etc, and Google can build up a very good profile of how people respond to your site on a keyword-level basis.

Would they reward site performance with better rankings? I think a better question would be: would they rank a poor site higher than a site with good usage data on the basis of other Algorithmical factors like links etc? Surely there's more value in the former, whether or not it's fully realised at this time.

Only supposition. No science.
Reply | Reply with quote | Quote
 
 
0 # Marjory 2009-01-15 09:44
I completely agree with you and I think you've really laid out all the issues in a really clear fashion.

There is one use of "bounce rate" that I was wondering if we should completely discount and that's Google's use of it to evaluate the effectiveness of their own site. I assume they use some kind of analytics program on their own site to evaluate user experience, etc.

Is there any way that you can think of that bounce rate (as in bounce rate from the SERP page not the page that the link on the SERP page leads to) as an indicator of site performance could be used a metric in the ranking algorithm?

(I have to confess - I have not read all the articles in the bottom of the post yet so if you want me to do that first professor and then come back and ask mindless questions, I will - just hoping I could cheat). :whistle:
Reply | Reply with quote | Quote
 
 
0 # Mike Wilton 2009-01-15 09:58
I know I was riding you about this post all week, but for good reason. Its a great post and you had some great data to back your points. I really hope this is enough to silence this subject, at least for now. "These aren't the ranking factors we're looking for. You can go about your business. Move along..."
Reply | Reply with quote | Quote
 
 
0 # Dave 2009-01-15 10:14
@carps – they ‘could’ certainly understand things better on a site to site level, but ultimately there are not enough sites using it to get global data. Without 90%+ saturation they would once again be left with incomplete data and potential noise. There are even patents relating to toolbar data, email, readers, favourites and more… this still doesn’t give the complete picture.

And once more, the researchers haven’t even gotten that far (large scale) as they are still struggling to assign value to behavioural signals in the form of engagement and perceived satisfaction of said signals. Add to that user SERP clicking bias and click-spam, it becomes even more problematic.

I believe (tho some IR peeps don’t) that there is hope for these signals, but moreso in a personalized environment… the debate continues (more studies come out, I shall report here).


@Marjory – actually click data (interactions with the SERPs) and query log analysis (data from analyzing queries made) are looked at a lot by researchers in this area. But the problem (as mentioned above) is finding truly actionable data. Different people have different knowledge level of a topic thuse different click behaviour. Different demographics (age, ethnicity) show different clicking patterns on the SERPs.. and so on. A single approach to human behaviour simply doesn’t give a clear signal as we are a diverse bunch with varied behaviours.

So YES I do believe there is merit in such signals, but it seems the technology isn’t there just yet.

Come back next week for my little journey into a far more likely scenario on Google’s behavioural approaches :0)
Reply | Reply with quote | Quote
 
 
-1 # DBL 2009-01-15 13:06
What's a bounce rate?
Reply | Reply with quote | Quote
 
 
0 # Dave 2009-01-15 13:48
I fixed the comments DBL...

While different applications have different calculations, it can mean;

A. people that visit only one page
B. people that only stay for a short period (5 seconds or less)

See Avinash's post on Bounce Rates (www.kaushik.net/avinash/2007/08/standard-metrics-revisited-3-bounce-rate.html)

That should give U the idea...
Reply | Reply with quote | Quote
 
 
0 # AJ Kohn 2009-01-15 15:50
Never say never.

I don't think Bounce Rate per se is a signal but something called pogosticking is likely a signal, though perhaps a weak one.

Many of the problems you cite seem small if you're looking at aggregate data.

Here's an excerpt from my Search Pogosticking and SEO post on Blind Five Year Old.

http://www.blindfiveyearold.com/search-pogosticking-and-seo

A user is presented with search results based on a specific query. The engine captures what result you click on and whether you return to that SERP and click on subsequent results and/or refine your query. (They could even conceivably determine the time between each click as a proxy for satisfaction with that result. This would reduce the chances of penalizing results that did deliver value.) The information can be aggregated for each query and compared to average pogosticking behavior by SERP rank.

So, let
Reply | Reply with quote | Quote
 
 
+1 # Glenn Murray 2009-01-15 17:32
Heya Dave!

As usual, great post, and excellent research base. I've got some tasty reading ahead of me!

For what it's worth, I agree with you that click-thru data and bounce rate are too noisy NOW.

But for what it's worth, I'm with Richard. Google has a pretty good track-record of filtering out noise (e.g. search spam of all sorts). I'm sure they'll come up with something to deal with this relatively effectively. Don't ask me how; my poor little brain can't cope with all that sophistimicated maths!

Also, I think SERP click-thru is a relatively minor signal, even if noise was filtered out. Logically there's only so much that can be gleaned about the perceived relevance of a page when Google has already skewed that perception, and there's only 10 or so results to compare (and most people look at only 2-5). Furthermore, there's little to be gained about the QUALITY of that page, because the reader can only see a snippet.

In any case, those are just two visitor signals (SERP click-thrus and bounce rate). What about the following? (I know you mentioned some of these factors in your intro, and your post is about click-thru and bounce rate, but the below signals need to be considered too.)

- How many people visit your site
- How those people arrive (i.e. non-SERP referrers)
- How long they stay
- How often they come back
- How many pages they visit
- What pages they spend the longest on
- Whether they comment
- Whether they subscribe to your blog feed
- What pages they bookmark
- What keywords they use in their bookmarks
- Who they share their bookmarks with (and who those people share their bookmarks with)

No single signal (or two signals) is without problems. But together, all these signals can give Google some very rich information.

I agree that if Google was relying just on Google Analytics to gather its data, it would need saturation before it could draw any reliable conclusions. But it's NOT relying just on GA. It also has all of the following at its disposal.

- SearchWiki
- Google Toolbar
- iGoogle
- Web History
- Google Bookmarks
- Google Desktop
- FriendConnect
- AdSense
- Google Reader
- FeedBurner
- Gmail

All of these tools and services feed visitor behaviour data back to Google.

I'm certainly not saying they can filter out the noise well enough YET to weight user signals highly. But they can definitely deduce something.

In fact, some of the engines, by their own admission, are already factoring in user signals. (See Cedric Dupont's (Google) comments at http://searchengineland.com/google-searchwiki-launches-15561: "We're always looking at user data as a signal." And see Tim Mayer's (Yahoo) comments at http://www.marketingpilgrim.com/2007/06/smx-notes-personalized-search-fear-or-not.html: "We use signals from those [Del.icio.us social bookmarking] pages to increase diversity." (Yahoo owns Del.icio.us.))

Anyway, thanks for a GREAT post and for your thorough research. Thanks also for sharing your list of readings. You're a champion. I don't care what @yetanotherben says about you behind your back! ;-)
Reply | Reply with quote | Quote
 
 
0 # John 2009-01-16 06:31
I just thought this WebProNews article will be very useful and relevant on this discussion.
http://www.webpronews.com/topnews/2008/11/17/seo-about-to-get-turned-on-its-ear
Reply | Reply with quote | Quote
 
 
0 # Glenn Murray 2009-01-16 06:51
It's an interesting article. (I wrote about it a while back on my blog - http://www.divinewrite.com/blog/seo/bruce-clay-buzz-king-is-pulling-our-legs/ ). But it's kinda looking at everything from a slightly different angle. ie. Bruce is talking about the search engines adapting to each individual based on their behavior and preferences. What we're talking about here is the search engines altering the SERPs for everyone (well, before personalization kicks in).
Reply | Reply with quote | Quote
 
 
0 # Richard Fergie 2009-01-16 06:58
Ok, I've done some more reading now. The majority of the papers from your list that I looked at seemed to think that including implicit feedback data did improve search result quality.

The issues seem to be filtering out the noise and spam. I guess filtering noise is much harder in real life than in the studies but I don't think I know enough to comment on this.

Spam clicks on the other hand are already dealt with very well by Google in a different context: AdWords. Without effective spam click filtration the AdWords business model just doesn't work. googleblog.blogspot.com/pdf/Tuzhilin_Report.pdf contains an evaluation of the effectiveness of AdWords click fraud detection and an overview of the methods used.
Reply | Reply with quote | Quote
 
 
-1 # Dave 2009-01-16 08:55
@Glen... let me get back on that...lol... might be a post in iteself... I do give ya kudos on the pgogsticking, but the confirmed what I felt last night with some IR peeps, that the method is still weak and not a lot of IR peeps have faith in it - but they are some interesting concepts.

As for your list, we already talked about most of those here... I kept my list limited as well, I've aready talked about it here 1/2 dozen times... lol

And the different data collection points, they do look at a variety of signals that come from Application Focus, which we've also talked about more than a few times here (want posts let me know) - there are certainly some interesting angles, we simply don't have data on the reach/market share for them - Ud need heay saturation to really find valuable signals...


@John - unfortunately the WebPro and Bruce Clay stuff is supposition besed on NOTHING... show me some research peeps. That's what I want from SEOs in 2009 - work for it.

@Richard - you are correct and that is what I said in the post - it has shown promise in segmented testing (not large scale real word). While some improvements have been made, intent an spam are still problematic. It works well in an enterprise search setting where manipulation is less likely.

As Glen eluded to, I do believe there is value in a personalized search setting. BUT and it's a big one, adapting for ONLY personalized simply isnt that easy from what some of my IR peeps are saying (and she's taking her PHD... so not a rookied with 10 years UNI on this stuff)

For me there is a more likely scenario out there (for Google personalized search) and I am working on a post for next week about that. (yes, I am approaching the PRO behavioral, just not the one we've been talking about up til now)

Thanks from dropping by gang, now to get my day started. We will be looking at more of this stuff next week, just a different angle, so we can chat more then :woohoo:
Reply | Reply with quote | Quote
 
 
+1 # AJ Kohn 2009-01-16 10:45
Yahoo! is clearly looking at pogosticking behavior via their patent application and Google has had personalized, SERP rank tracking in place for a number of years (as i described in my link previously).

I doubt very highly that Google is using GAnalytics data. They can get this information in other ways that don't pose such an enormous risk. That said, the benchmarks product within GAnalytics might be something they look at and match up against their own site taxonomy, which I believe they do have on some level.

SearchWiki is the more interesting avenue right now IMO. At its core, it provides a human feedback mechanism on SERPs. The algorithm needs a human tutor and SearchWiki turns us into a massive army of mechanical turks.

www.blindfiveyearold.com/searchwiki-turns-you-into-free-mechanical-turk

Again, remember that we're talking 7 billion US searches in a month on Google. It only takes a small amount of usage to obtain a stream of intelligent (human) ranking via SearchWiki.

It's not about the rankings of any individual but whether the human ranking of that SERP deviated statistically from what the algorithm produced.
Reply | Reply with quote | Quote
 
 
0 # Nick Stamoulis 2009-01-16 10:29
There is a lot of back and forth on this. Personally in the future I think that this will be a factor in rankings.
Reply | Reply with quote | Quote
 
 
0 # Dave 2009-01-16 12:27
@AK John - well, Yahoo also put out patents relating to Personalized PageRank an HarmonicRank... sooooo... One learns quick that patents are mearly covering ones ass... not always a sign of actual implementation.

As far as Goog, Search Wiki signals used in a non-personalized enviro would be MASSIVELY spamm-able, regardless of what Marissa says about 'some day' there are many problems inherent there. Personalized search? Sure... it's an explicit signal of worth. To many people underestimate the problems associated with adversarial IR - tricky business

@Nick - there is considerable debate in the IR world as well. Obiously behavioral metrics will play into the evolution, we're just not sure what flavour (approach) - I do agree with you there...
Reply | Reply with quote | Quote
 
 
0 # Lea de Groot 2009-01-17 01:12
When combined with personalisation its going to be much more useful:
Take my logged in personalisation , put me in a group of other people who seem to have similar interests, and prefer results for me that others in my group are bouncing less on.
There, was that so hard?
So, I just have to hope that my usage pattern doesn't make me look like a 3rd world paid-to-click person :sad:
*sigh*
Reply | Reply with quote | Quote
 
 
0 # Dave 2009-01-17 20:23
That's interesting U say 'groups' as much of the personalization does work that way as it can be to resource intensive to go on a singular level.

I have a post Monday on 'Personalized PageRank' be sure to tune in :0)
Reply | Reply with quote | Quote
 
 
+1 # fantomaster 2009-01-18 18:27
Just to chime in on a couple of points seeing that you mention me explicitly, heh.
For one, if you want to go for manipulation of SERPs via behavioral signals (I'll leave it to Matt Cutts and his minions to term it "spam" or, worse, "black hat" in Matts' current favorite synonymizing tactics "black hat = felons" - think: crackers, malware infusers etc.), you'll obviously have to scale your activities.
Claiming, like some patents do, that the overall impact of such "noise" (which, technically, won't actually even be that: manipulation, yes, but not "noise" in the sense of "chaotic signals" will be negligible is quite ludicrous, to put it mildly: deploy a and scalable network of SEO Surfbots for all the relevant search terms (short tail and long tail alike) and let them do their good work automatically 24/7.
Once rolled out, all you'll have to do is feed them with lists of search phrases and target sites, that's all. Do it on a couple of mainframes, and where does that leave all those super duper protection algos? Yes, it's brute force, but then so is spam and scalable IM anyway.

As for tying personalized search into related/similar groups to cut on bandwidth and computing power: sure, that's what SEO Surbot Nets are all about, really: mimicking the industry-typical customer will create those very patterns that can function as false positives to dominate results.

Personally, I, too, don't buy into the hype that the major engines are actually using bounce rates etc. in any major way currently. But I won't put it past them that they'll go for it one day if only to cut some corners. After all, doing this is a whole lot cheaper than developing highly targeted verticals with loads of human editors to monitor results and safeguard quality...

As for that old carrot of "Google being great on catching spam", my sole response is "bah". (Check results for "cheap viagra" or similar to see for yourself - and lots of those SERPs are very probably hand jobs, too...)
Reply | Reply with quote | Quote
 
 
0 # Dave 2009-01-18 21:59
From my understanding a 'bounce' is recorded when someone visits a web page but then doesn't visit another page on the website after that.

This behaviour supposedly indicated that the user was not happy with the page they arrived on, so decides to leave. This interpretation is not valid for all web pages. I can think of plenty of times when I visit one page directly from Google, get the information I want, and leave again.

It seems to me that bounce rate is an old way of thinking about the web when people had homepages to attract people, who would then move on from there. It's much more common now to deep link into a website, meaning the user does not necessarily have to move on from that page. This would result in that page having a high bounce rate, despite the fact that it's meeting people's needs.
Reply | Reply with quote | Quote
 
 
0 # Dave 2009-01-18 23:23
whoah... 2 Dave's - we're talking echo chamber here soon...

@Fanto just had to leave Matt love notes didn't ya? I guess that's why it's 'adversarial IR' huh?

I think that's why some IR peeps that feel this course is a dead horse, if may be lyrical at this late hour. No one has done more than show 'my pretty idea' in a closed environment (often with woefully poor subject diversity). The spam element solicits a cry for more funding.

Now the 'noise' part, is actually more about establishing intent and satisfaction ie; perceived signals vary wildly and assertions are unclear from demographic to demographic.

Thus between these two relatively large elephants in the room, one wonders if there is hope ultimately.... It would suck if it's a ghost, I've put so much time in on it...lol :o


@Dave - hi there, Dave here... (told U the echo chamber was coming)

There are two main schools which are; visiting one page and/or time one page (say, less than 5 seconds).

You raise more valid points. If Google is doing it's job, for example on an informational query, I should get what I was after and be out of there. If I had to dig 3 pages deeper to find that which I was after, how does this show satisfaction? In some cases this is a positive, in others not at all - these are the types of 'noisy' signals I mentioned above (the list goes on and on).

I do agree with your assessment, though it can be situational and query dependent or not.

.... time will tell.
Reply | Reply with quote | Quote
 
 
0 # currency trade field 2009-05-15 04:57
Thanks for your comments. Google would be able to work out bounce rates via both Analytic and their site itself.for example lets assume I search for
Reply | Reply with quote | Quote
 
 
-1 # http://www.greececasino.com/ 2009-12-15 09:56
I guess filtering noise is much harder in real life than in the studies but I don't think I know enough to comment on this.tying personalized search into related/similar groups to cut on bandwidth and computing power: sure, that's what SEO Surbot Nets are all about, really: mimicking the industry-typical customer will create those very patterns that can function as false positives to dominate results.
Reply | Reply with quote | Quote
 

Add comment


Security code
Refresh

Search the Site

SEO Training

Tools of the Trade

Banner
Banner
Banner

On Twitter

Follow me on Twitter

Site Designed by Verve Developments.