A web analyst needs to develop and establish controls which can then be used to maintain an acceptable level of data quality. The credibility of the conclusions drawn from analyses depends heavily on the quality of the data contained in them. Data reliability is therefore an important pre-requisite in the field of web analytics. If there is the slightest doubt relating to the quality of data, questions may be raised about the work involved in analytics. Data quality control is therefore essential, and is possible thanks to one of the following methods:

• Daily email report summary

The email report summary can be used as a data quality control tool. Recipients receive an email everyday which provides them with information on their main web traffic data. Below is an example taken from the email report summary sent on the 26^th October 2010:

It allows you to react to any possible tagging problems which may have been encountered the previous day.

The report summary for a specific page can be configured by going to: Tools > Customisation > Email reports.

I recommend that you add your site’s homepage, conversion pages and any other important pages which are tracked on your site to the report summary.

• Weekly control table

This control table ensures that there have been no tracking problems for the week which has just ended. The table provides you with information on the evolution of the main variables and items for the week which has just ended. Evolution curves allow you to spot decreases in traffic volume or any other unexplained decreases which may not have been detected by the alert systems and the daily email report summary.

The following list shows the main variables which the evolution curves examine:

• Number of visits
• Number of page views
• Number of conversions
• Number of visits by traffic source
• Number of visits to the main sections of a site
• Number of visits to the main pages of a site

Each Monday, you need to make sure that the most important data for the previous week has been collected. This data is the key information which is then to be used for data quality control.

• Real time alerts

If an online publication, or any other modification, changes the way in which the tag works, this will result in a decrease in the value of traffic volume variables such as visits or page views. You should then set alerts for whenever your site experiences low levels of web traffic. Such alerts must be triggered immediately.

Analyzer NX provides users with an alert system which can be configured in the interface.

The alert is sent by email. Web analysts must check if the tool confirms the event which has been highlighted in the alert. They must then check if such an event is due to a problem with the site’s tagging.

Web analysts then need to alert their production manager or web agency immediately if the tag is no longer functioning correctly, and point out that it must be corrected as soon as possible.

These different alerts, coupled with the reactivity of the different teams reduce any period of inactivity during which no measurements take place.

By following this procedure you will avoid having to announce the bad news to your superiors that it is not possible to carry out a particular analysis because of a tagging problem. If you have other methods in place which are used to control data quality, or, if thanks to this control you have been able to detect certain errors, then please feel free to share your experience with us!