It happens frequently that programs such as Google Analytics or Lloogg produce statistics in defect respect to the ones produced by your server.

Photo Credit: AnaBGD

Fake visitors

The reason is very simple: the programs of web analytics that monitor the accesses reading the log files of the web server (as awstats for instance) erroneously evaluate like visits many contacts that are not.

Here a list of false positive that can wreck the web stats:

XML engines

When you submit your feed into XML aggregators, a bot check your feed for updates frequently. Feedburner, for exaple, works this way: more your site is up to date, more it comes frequently producing a lot of access logs which could be confused with real visitors, but it are not.

Spiders

It’s the same for search engines spiders that index your web pages. Some times it are so intrusive that can bring your server to crash.

Trackback spammers

This is one of the most common causes of false positive.

If your blog support trackback it is subject to be spammed. In fact spam bots continuously call the trackback script trying to insert some links.

Some of these bot are able even to modify its referring IP every time they call the script, and could therefore be seen as various visitors.

Comments spam

It is the same for the comments; bot continuously try to insert fake comments with links inside calling the script and generating a new log each time. Each log line can be confused with a visitor.

How to get perfect web statistics

In few words, it is not possible to get an affordable idea about web traffic using just a log analytics application.

It would be better to use an outside application, which makes use of JavaScript to track the visits on the content pages (I recommend LLOOGG, i have some invitations), and make comparison with the stats produced by your web server.

In this way attempts of spam on trackback or comments script will be ignored, and the visit counter will increase just when JavaScript will be executed on the content pages.

In any case you will never have perfect stats, but you will be able to have a more reliable idea about the real traffic of your blog.

I started testing it on a couple of web sites after obtaining an invitation as beta-tester from antirez (AKA Salvatore Sanfilippo, cofounder of Merzia together with Fabio Pitrola).

There is an essential difference between a normal web analytics program (like GAnalytics, for example) and LLOOGG: it allows you to track in real time accesses and navigation of the users on your web site and provide a lot of useful information like:

• Visitor origin
• Keywords
• Originating country
• Graphic resolution of the client
• Browser and OS
• navigation path

Nevertheless, it doesn’t finish here. These data are already present (more or less explicitly) into the logs of every web server. These are the winning points of LLOOGG:

• It is a different, but complementary, service respect the traditional web analytics that uses log files to generate stats
• It provides information in real time using an Ajax web interface. A browser is all you need.
• It improves and makes more readable the data already present into the httpd logs
• It let you to correlate between the each log related to a single user thanks to a colored marked

It’s sufficient to insert a JavaScript code, which will detect the user’s visit into each content page that you want to track and the visitor information will start to fall into the web Ajax interface. It’s seems a kind of Matrix effect. Look:

As you can this visitor landed on this page searching for “voipstunt” via Google.co.uk and it come from United Kingdom. There is also his monitor resolution and browser type/version. This information block will fall down from top each time a user accesses a page. Look on the grey marker: every user is marked by a colour, which allows you to identify it in the long list of accesses in order to follow his navigation.

This user is reading several pages:

Reading from top to the bottom:

1. he entered directly into the VoIPStunt article page
2. he clicked to the VoIP category link
3. he read the Gizmo Project article
4. he went back to the VoIP category page
5. he is now reading the guide about mobile VoIP installation.

Even if geeks will enjoy the idea of a web analytics in a Matrix style, imho LLOOGG strongest point is about web marketing: in fact it allows to monitor the efficiency on the immediate of social news and social networking that web usually use to spread our contents.

In this way I discovered that also the rejected news posted on Slashdot could produce an interesting traffic amount thanks to the approval tail any way available for logged users.

Log blocks obviously will populate an historical DB, which will be used to generate chart and stats. It’s still not clear exactly what kind of features will be available for the PRO account and what will be available for the free one. Anyway, this is the list of features currently available for beta testers:

• Maximum number of logs shown in the Ajax interface (LIFO list)
• Pause/restart monitoring buttons
• Top Referrers
• Top Searches
• Monthly/daily/hourly charts and stats
• Browser Stats

LLOOGG is currently available only on invitation. If you want an invitation just leave your email using the comment form.