Page 8 - Experiment to test user behavior – please participate

#4 for me. I am surprised in a way. I didn't expect this much change with the few participants we had JUST for clicks. I expected maybe 6-7-8 but not 2-4.

I also have a little feedack on my research and will soon be putting out an article on it. I have it doesn't take that many users to drastically increase your rankings. It is not so much the numbers what they do on the site. If they spend a long time at the site and visit many pages, spending a long while at each it will give you a great boost. This is of course with analytics installed.

I also have evidence to support my believe that long visits will increase many of your positions.

On one of my sites that ranks in a very competitive field. It ranked the same position for about 7 months or so. I have not worked on this for a long time, I don't have any time for it, the rankings haven't increased at all and the backlinks haven't. Yet when I used test users to spend time at the site after doing the google search it jumped up 45 positions and is now gaining traffic.
I have also been watching some of my phrases for my seo blog. I notice they rise when the visits to that page are long and low bounce rate and when the bounce rate rises(about 65+) and the visit length gets very low I began losing the position.

I am still working on some more test experiments. But behavior tracking will play a major role from now on, don't under-estimate it.
If you have any questions regarding my research let me know.

Although I quite enjoy your efforts for some new tests I'm not sure about the long visits part. Yes you can track tons of things but Google bot doesn't hang around and determine length of visitor stays.... They could track clicks to the web site though.

The google toolbar would give them info for time spent on site.

Google actually does not only measure the lengths of visits if Google Analytics is enabled, it also tells you as the webmaster about it. True, only a fraction of all sites may have analytics in use, but this information definitely is available.

Proof positive they do if Analytics is indeed installed. I can't proof they can track it any other way but I can tell you they do if Analytics is installed. I improved one of my pages, the visit length went from average 2 minutes to about 30 minutes and a good deal at 60 with same bounce rate. I saw a great increase in rankings even though the page was near dead.

Its currently at #4 for me!

done

#4 for me as well

Its back to five on mine now. This thread has not been up the top for days now. I am guessing that what may now be happening is that when the traffic dies the page will return to its previous position over time and google 'normal' ranking will reassert itself.

Since reading the article http://www.goupstate.com/apps/pbcs.dll/article?AID=/20070603/ZNYT01/706030354/1051/NEWS01

I Think the rise in rankings due to clicks and google's effort to identify when 'fresh' content is desirable may be part of the same algo change.

My reasoning goes like this. If there is a sudden large increase in google for a particular search term than google looks for pages that have 'fresh' content on their site with regard to this search. The criteria it then applies to re-ranking serps is based on clicks and bounce rate.

This would allow the algo to re-order current/new topics in order of what people wish to find when they are searching that term. I imagine this would be quite easy to incorperate into the algo and would give the results suggested that google wants from the info in the article. Thus it is not so much 'freshness' that is important but 'current relevance' that is being used in ranking. This would mean any page that suddenly sees a large increase in traffic could expect its rankings to rise.

I suggest a simple test that may offer support for this for interested people. For you city type the citys name in google and take note (or a screen shot of page 1 & 2) ( a topic like health research would also work i think) than when a major news story breaks about your city (issue) check to see if pages that have relevant, fresh content suddenly start to appear in SERPs for you citys (issue) query. See if there rankings increase over the time that the story is current. If any one does this and sees a change let us know. Than I would expect that when the stroy stops generating clicks the page will fall back in serps again.

Once again thanks for those who have participated in this experiment. It does really seem that google has beeen making some fairly significant changes and we are just starting to learn their effects....

Gazz, for this I have a proof like hell, but I was stupid because I didn't make a screenshot before (particularly because I didn't know about QDF's), so you will have to believe me.

Check out the search "bush in prague". I know that before a week the SERPs for this keywords were only articles about the Bush's previous visit to Prague's NATO conference in 2002.
Now, there are news articles which are only few hours old... His speech was in SERPs only few minutes after the speech was published: http://www.state.gov/p/eur/rls/rm/85995.htm

I don't know how they do it, but they do it well.

#4 india..

Found a drop in rankings. It's now 8 in India.

May be just a usual flux coz it was 5 yesterday

Thanks Gazzy and Diffee for conducting such a great experiment. Now we have rock solid proof to claim that Google do consider user behavior while ranking sites.

Done.

Guys, its showing at #2 in India....

Anyways, good efforts...

Hi gazza,

I think your experiment is a great idea and I also believe Google monitor click-throughs from their organic listings - it makes sense to know where their audience is clicking on their results pages.

Just a few thoughts though :-

- the keyword you are using for this experiment produces a very small results set - 859 results returned when I did the search (you were #7 by the way, google dot com, Thai IP address searching). I wonder if this theory (if proven) is amplified on smaller result sets? Would it apply to result sets with 1 million plus results? My own feelings are that that's where link authority would matter much more.

- what about sites that don't use Google Analytics? Yes they can measure the length of stay from users who have the Google Toolbar installed, but I think it's not conclusive enough to show a site has "sticky" information - you'd need near-universal metrics (i.e. all sites using same exact same metrics) to compare like-for-like.

No matter, I think your theory has some truth to it, simply because it makes sense for Google to show the most popularly clicked results closer to the top (e.g. if a result on page 2 was the 8th most clicked result, you'd surely want that higher up the rankings).