Comments on: Record-Breakers

By: Jonathan Lewis

Jonathan Lewis — Sat, 17 Apr 2010 11:22:51 +0000

Gary, I'd like to start by thanking all those kind people who encouraged me to believe the 20M+ figure - but I think your explanation is rather more likely. However the Nov 2009 figures (the newest I can find) for Wordpress show 1.4 billion pageviews across the board; but a slightly coincidental 25 million "active posts".

By: Nigel

Nigel — Tue, 13 Apr 2010 10:31:39 +0000

Sokrates:

Google only shows max. 1000 results, even if there are many more matching pages.

By: Sokrates

Sokrates — Tue, 13 Apr 2010 07:55:55 +0000

“accuracy in counting websites / webhits” that’s a great joke !

have a look at google’s accuracy for example:

a simple search for “Sokrates”
http://www.google.com/search?hl=en&q=sokrates&sourceid=navclient-ff&rlz=1B3GGGL_deDE285DE288&ie=UTF-8
yields “… Results 1 – 10 of about 1,720,000 for sokrates. (0.05 seconds) ”

everyone is very impressed by this.

However, when you start to look more deeply into it, you see that “about 1,720,000″ in reality means “at least 900″:
http://www.google.com/search?hl=en&rlz=1B3GGGL_deDE285DE288&q=sokrates&start=10&sa=N
gives you ” Results 11 – 20 of about …”,
http://www.google.com/search?hl=en&rlz=1B3GGGL_deDE285DE288&q=sokrates&start=890&sa=N
gives you “Results 891 – 900 of about …”
which is then end !

http://www.google.com/search?hl=en&rlz=1B3GGGL_deDE285DE288&q=sokrates&start=999&sa=N
then gives you
”
Your search – sokrates – did not match any documents.

Suggestions:

* Try different keywords.
“

By: Gary

Gary — Mon, 12 Apr 2010 22:39:34 +0000

If Compete are in the habit of counting “otn.oracle.com” in the same bucket as “forums.oracle.com”, I suspect the result actually means that all the buckets under “wordpress.com” makes it rank as the “33 most visited website in the US”.

By: Andrea

Andrea — Mon, 12 Apr 2010 21:08:14 +0000

I second Kent’s theory… I was here at least for another 20K visits :-P

By: Eslam

Eslam — Mon, 12 Apr 2010 20:07:33 +0000

No wonder why that, because your blog is always active and full of information

By: kent

kent — Mon, 12 Apr 2010 15:14:51 +0000

Id believe compete.com, as I reckon Ive been here at least 20K times on my own in Feb ;)

By: Timur Akhmadeev

Timur Akhmadeev — Mon, 12 Apr 2010 13:11:41 +0000

From this: >Compete’s clickstream data are collected from a *2,000,000 member* panel of US Internet users (about a 1% sample), using diverse sources. Using a rigorous statistical normalization methodology, Compete creates *precise projections* of the behavior of the entire US Internet browser population on monthly and weekly basis. it seems like a bug. Totally.

By: Nigel

Nigel — Mon, 12 Apr 2010 12:01:33 +0000

Judging by a few reviews I have read of comlete.com’s accuracy, I think that you could arrive at similarly “useful” stats, as follows:

Think of a number.
Add fifty billion.
Subtract your age in whole years.
Add the number of molecules of water in the Pacific Ocean.
Add another number… the bigger the better.
That’s it.

I like Scratchpad, but I doubt that it’s acing Amazon, eBay, Fox News, IMDB, iTunes, Wiki, Facebook, Twitter, Flickr, michaeljackson.com, and the top 23 guns/porn sites in the States.

Sorry to be a party pooper!

By: Tony Sleight

Tony Sleight — Mon, 12 Apr 2010 11:43:44 +0000

It doesn’t surprise me that two so vastly different values are published. It depends upon how you class a visit. If you class a visit as what used to be called ‘hits’, then each item that is downloaded from your website would be classed as a hit, that is, each image, page, file, link etc. So over a month 45,000 visitors could have downloaded 500 items from your site giving you 2,250,000 hits! If you class a visit as a session, then you need to know when a session ends. Generally, this is classed as ending after 30 minutes of inactivity. Therefore, the number of user sessions is going to be far less than the number of hits. So, if wordpress count sessions, and compete.com count hits, then both are right. You choose the one who gives you the data you require to measure your website performance. A bit like execution plans come to think of it! Both could give those I/O values for the same query! Just depends how you process the query.