SEO Blog - Internet marketing news and views  

Behavioural metrics and search engines

Written by David Harry   
Friday, 02 January 2009 13:21

The debate continues

Well it seems the whole discussion of bounce rates and SEO continues… and as predicted, it’s picking up steam. I think it’s important and worth stating once more; I do believe in behavioural signals and their application just not the current representations floating in the sphere.

We’ve talked about them here on the trail since 2007 and here’s a list of posts on the various aspects;

Beware: Google is watching you
Query analysis confirmed as Google ranking signal
Taking off my tin foil hat
Ranking through behavioural metrics
You’re not paranoid, someone IS watching you
What every SEO should know about personalized search
Personalization through application focus
Personalized search and behavioural metrics
the Art of of personalized search optimization
Yahoo personalized PageRank

And here are some patents that I didn’t get around to covering;


Using search trails to provide enhanced search interaction
Personalization of web page search rankings
Click through log mining
Accounting for behavioral variability in web search
Automated analysis of user search behaviour


Search pogosticking benchmarks


Personalized Network Searching


Why are we fixated on bounce rates?

I think the most troubling part of this is how SEOs are so fixated with bounce rates. There are more than a few ways (implicit and explicit) that search engineers can find behavioural metrics from which to glean signals. Why is it that the community has latched onto one of the weaker ones (bounce rates)?

To review, here are potential avenues of data to track;

  1. Query history (search history)
  2. SERP interaction (selections and bounce rates)
  3. User scrolling behaviour (on selected page);
  4. User document printing behaviour;
  5. Adding a document to favourites (bookmarks);
  6. Data from different application (application focus – IM, Email);
  7. Surfing habits (frequency and time of day)
  8. Interactions with advertising
  9. Demographic and geographic
  10. Explicit editorial (such as SearchWiki voting)
  11. and closing a window.

As you can see there are many approaches to analysis of behavioural signals. One wonders why only one has been touted by the SEO community; bounce rates. Is this ignorance? Or the one they thought could be manipulated the best? Who knows.. but to me it really does a disservice to the community to have such limited speculation and reporting.

…for the record? I’ve never seen ‘time on page’ mentioned in any patent/paper, so let’s not even go there, ok?


It’s a personal thing

Ultimately we have to concede that search engines are quite interested in this area, but we simply must consider the problems such methods would have in the organic SERPs. It is far more suited to a personalized search environment than anything as it would be kinda hard to spam yourself.

This is where folks seem to not be understanding my position on this; I have been writing and talking about it for a looooooong time, just not so sure about it’s efficacy or implementations in the day to day ranking mechanisms. And calling it a ‘new’ or ‘major’ signal is nothing short of nonsensical.

And we’ve even discussed ways one can adapt their SEO programs for behavioural/personalized search here;

Another problem in getting a clean signal (in a non-personalized setting) is the assertion that there are enough data points to make this all effective and that is debatable IMO.


Where’s the beef?

For me the problem is when SEOs start running around trying to position themselves as experts (or get you on mailing lists) using this talk of behavioural metrics. Not to be tooting my own friggen horn, but I have talked about this (in great detail) for more than a year – BUT I am not offering ebooks nor making predictions about things that have been around for quite some time now. That’s what’s really eating me… and it has already happened.

For example is this offering; ‘Sticky SEO’(the link is 'followed' I ain't a total dick)

Which has statements like; “Please note: this book is not just a bunch of theory. It is a practical guide that offers 6 exclusive strategies to master this new "usefulness algorithm".” Or; “What should you do to score big with the new algorithms?” - - That just smacks a bit of self rightiousness (and hype to get peeps signed up).

And while Fantomaster did a great job of approaching the topic in a sensible matter, he does bring the prospect of spamming said metrics with; SEO Bounce Rates, Behavioral Metrics and the Birth of SEO Surfbot Nets

Gee… pretty sure the search peeps saw that coming and thus Matt (Cutts) and John (Mueller) have recently poo poo’d the inference by SEOs of bounce rates as IR signals. Now if we take that at face value or not, we should at least consider it.


Have any questions?

Since this debate is growing, as expected, I now leave the floor open. I have done as much as I can to try and bring some reality to this and unfortunately the masses are more interested in ‘the Big Story’ and not discussing the reality of the problems facing search engineers with such methods. Feel free to get in touch to discuss further... because that's what is needed, not hyperbole.

On Monday we’re going to look at a TON of resources that the fastidious SEO can use to learn more about Indexing and Retrieval and technical aspects relating to them. If you really want to be better at understanding search engines… I suggest you tune in then.


Peace ya’ll… it’s the love of the art that keeps me rolling…please take my rant as that.

See this post for more details on bounce rates and behavioral metrics

Search the Site

SEO Training

Tools of the Trade


On Twitter

Follow me on Twitter

Site Designed by Verve Developments.