JUST THE FACTS thread - Google's algorithm change

This for compiling everything in this thread Mike. I thought I'd share my experiment here...

Something that content farms often do is abuse Adsense. I did a little test to see if I could find some sort of correlation between the use of Adsense and some of the sites that were most negatively affected by the algo change. I wanted to see if sites that linked out more to ads than real sites got hit for it.

"...parts of our system encourage links to good sites."

-Matt Cutts

"...good pages seldom point to bad ones"

-http://www.vldb.org/conf/2004/RS15P3.PDF (TrustRank)

The ads don't necessarily lead to bad sites. Quite contrarily, they often lead to authoritative pages. I bring this up simply to point out the fact that Google uses your outbound links as a factor when ranking your site. So, if content farms typically have a larger percentage of outbound links that are ads, it is plausible that Google might have started using this as a measurement in their algorithm.

I took a random sample of 5 article pages from each of these sites and measured their adsense links to OBL (outbound link) ratio as well their adsense to overall links ratio. Here's what I found:

(In order of most negatively effected by the update to least)

Suite101.com - Change: -94%

Ad Links/Total Links = 14.5%
Ad Links/Outbound Links = 54.14%

Answerbag.com - Change: -91%

Ad Links/Total Links = 11.72%
Ad Links/Outbound Links = 60%

Buzzle.com - Change: -85%

Ad Links/Total Links = 21.79%
Ad Links/Outbound Links = 92.42%

Wisegeek.com - Change: -77%

Ad Links/Total Links = 15.93%
Ad Links/Outbound Links = 67%

And here's what I found for Ehow...

Ad Links/Total Links = 10.28%
Ad Links/Outbound Links = 38.83%

The samples were relatively small (hardly empirical), and the data didn't quite correlate to my hypothesis as much as I had hoped. However, in the test I conducted, Ehow's Ad Links/Outbound Links ratio is noticeably lower than the other sites'. This leads me to believe that one of the ways Google tried to target content farms is by cracking down on sites who mostly link out with ads.

What do you guys think??

P.S. I couldn't get SEO Quake to measure the links on articlebase or ezine or I definitely would have used them

You are missing the point. They can track every user that is logged in (usually logged in because of Igoogle or gmail) using their search engine . Thats how they can give personalized search results. No GA is required and no toolbar. They cannot get your direct bounce rate but they can track that the users chose another link in the search results which is all they need to determine how long someone stayed at a previous link. Its practically the same as a bounce rate most of the time.

Servers can time stamp almost everything so its data they have right in front of them and its a top candidate for determining how useful a search result was. If you find what you are looking for after being served a link then you are not going to be coming back to google immediately to have another link served. I know I rarely stay at an ezineartilce article. I am right back clicking on the next link on Google.

Are you sure? Everyone gets a cookie when they do a Google search - if you click on one search result and then you are back on Google clicking on another link within 10 seconds then that's a bounce, surely?

Anyway, MILLIONS of people have the toolbar installed, that must give them a pretty good indication of visitor behaviour!

All Google needs for bounce rate is your IP address and to track clicks.

Did a searcher with IP 4.4.4 click a link in the SERPs, then IP 4.4.4 click another link in the SERPs just 10-20 seconds, for example, later?

And if you have the Googlebar or use Chrome, they can get tons more info.

You are right, Google can't always be trusted with the statements they make.

With the information you have given it doesn't really provide any indication of anything in relation to bounce rate. Maybe you have a more clear picture, but comparing 3 different sites, with different content and many other different variables, it is very hard to get any sort of solid conclusion.

Since your main site also has more backlinks (and presumably higher quality) than the autoblogs, by the same logic you could state that backlinks are also meaningless and Google is lying that backlinks matter.

We know this is not true, so why assume it is true for bounce rate or back button rate?

You have compared 6 variables between the two sites:
- uniqueness of content
- time on site
- number of backlinks
- quality of content
- niche
- bounce rate

There's literally hundreds, if not thousands, of other variables Google uses in its calculation. It is much more likely that some of those other variables explain the difference, and in this situation hold more weight than the 5 variables you have listed. In other words something else about your sites is holding it back and trumping all those other quality indicators.

If you study the other variables and compare them across a wider net of websites, use a regression analysis type approach you will find your answer.

The challenge can be realizing what those other variables are, which takes some good detective work, insightfulness and an ability to be open minded.