Analytics 2.0

Much more than just tagging

Google Analytics with new improvements

Google has announced a new set of improvements, including New Graphing tools, GA.js and six new languages.

New Graphing Tools: Now you are able to graph two metrics against each other and see how they correlate. This is a very useful feature since allows you to carry on a more advanced analysis that not only considers separated set of data but also how any action you may take affect your site traffic and conversion. Here are the instructions on how to access this new graphing feature.

New GA Graphing tool

Track actions in a correct maner: Google has launched ga.js page tag that allows you to take advantage of some new tracking functionality like action tracking (counted as actions instead as pageviews), ecommerce transactions in a more readable way and a special resource on the ga.js javascript for power users who want more control over Google Analytics tracking. Anyways you don’t need to install the new tracking code, if you have a website already been tracked by Google Analytics you can leave it as it is, if instead do you want to use the new one to take advantage of the new features you can also do it (more information here).

Six additional languages: Google has launched the Google Analytics version in Thai, Filipino, Indonesian, Czech, Hungarian, and Portuguese (Portugal). Something in my opinion really important and completely aligned with Google Analytics market strategy.

Post to Twitter

The lack of precision in the Data Panels

In a preview post we was talking about the importance of a good information source. We also mentioned two information sources based on user panels, Alexa and Comscore.The main advantage is that comScore provide their Panel Users with software that measures lot of demographic information like gender, age, income, etc.
With the above mention information they project the data to the total internet data universe, using statistical procedures to correct potential deviation of information.
Like in the rest users panel there are certain limitations regarding the information accuracy. The best known is that sites with low level of traffic are not properly measured because the sample is lower than necessary to accurate an acceptable level of certainty.

Analyzing data accuracy

When using a panel like Comscore it is desirable having a constant relationship between the real data and the estimation it provide us.
In order to test this hypothesis we use the data from an important brazilian website.

As you can see in the following charts the estimations carried on by Comscore are significantly different from the real data. In some cases when the real increases in Comscore estimation it decreases and vice versa.
.

Datos UOL

Datos MSN
If you take a look at the trends from Terra we see that the real traffic have a different pattern than the estimation which indicate a lack of accuracy in the estimations provided by Comscore..

Data Terra

On the other hand the data from a panel of users tend to have a lower variability –since you are taking the same group or cluster of users- than the entire population. In the mentioned example, the coefficients of variation (expressing the ratio between the standard and the average deviation) from the Comscore panel are considerably lower than real ones.

Datos Terra UOL MSN

This show us once again that the problem does not end with the capture of data (Analytics 2.0-much more than just tagging :-) ) but we must also be focused on the information quality and methods we use to produce it.

Post to Twitter

Twitter links powered by Tweet This v1.8.3, a WordPress plugin for Twitter.