Comments on: Errors of causation in web analytics http://actionable-analytics.com/2009/07/errors-of-causation-in-web-analytics/ Thoughts on Web Measurement & Optimisation - by Jonny Longden Tue, 03 Apr 2012 09:58:32 +0000 hourly 1 http://wordpress.org/?v=3.3.2 By: Justyn http://actionable-analytics.com/2009/07/errors-of-causation-in-web-analytics/comment-page-1/#comment-169 Justyn Thu, 07 Jan 2010 04:39:24 +0000 http://actionable-analytics.com/?p=89#comment-169 You're exactly right! I have a client who is using a static web page for email campaigns. Since a LARGE group of their clients use Outlook, the browser isn't passing any information about a referral to us. So the customer is getting the email, clicking an ad, and our analytics show a "direct" referral. You can imagine their relief when I explained the issue and their email campaigns were actually performing above standard. My solution is to have them put tagged links throughout their email campaign so it doesn't matter who referred them...I tell the analytics EXACTLY from where I got the click. Here's the tool I use to build those tags out: http://www.google.com/support/googleanalytics/bin/answer.py?answer=55578&hl=en Thanks for making people challenge assumptions... You’re exactly right! I have a client who is using a static web page for email campaigns. Since a LARGE group of their clients use Outlook, the browser isn’t passing any information about a referral to us. So the customer is getting the email, clicking an ad, and our analytics show a “direct” referral. You can imagine their relief when I explained the issue and their email campaigns were actually performing above standard.

My solution is to have them put tagged links throughout their email campaign so it doesn’t matter who referred them…I tell the analytics EXACTLY from where I got the click.

Here’s the tool I use to build those tags out: http://www.google.com/support/googleanalytics/bin/answer.py?answer=55578&hl=en

Thanks for making people challenge assumptions…

]]> By: Jonny http://actionable-analytics.com/2009/07/errors-of-causation-in-web-analytics/comment-page-1/#comment-55 Jonny Fri, 31 Jul 2009 15:54:52 +0000 http://actionable-analytics.com/?p=89#comment-55 Hi Christian, Good question. I think the answer is two-fold: 1 - If we are ever going to get close to <em>why</em> people behave the way they do then we need to utilise qualitative data as well as click-stream, and sometimes even instead of click-stream. The only sure-fire way to get the bottom of some of these things is to ask people. 2 - In some cases the reasons for behaviour are simply too complex to understand, or at least to draw out using data. Take buying a car; we can spend a lot of time and money trying to measure the exact detail of multi-channel buying behaviour, but at the end of the day this process may be completely unique for every single person. Sometimes we have to take what we have and make the most of it - assumptions guided by insight! Thanks for commenting Hi Christian,
Good question. I think the answer is two-fold:

1 – If we are ever going to get close to why people behave the way they do then we need to utilise qualitative data as well as click-stream, and sometimes even instead of click-stream. The only sure-fire way to get the bottom of some of these things is to ask people.
2 – In some cases the reasons for behaviour are simply too complex to understand, or at least to draw out using data. Take buying a car; we can spend a lot of time and money trying to measure the exact detail of multi-channel buying behaviour, but at the end of the day this process may be completely unique for every single person. Sometimes we have to take what we have and make the most of it – assumptions guided by insight!

Thanks for commenting

]]> By: Christian http://actionable-analytics.com/2009/07/errors-of-causation-in-web-analytics/comment-page-1/#comment-54 Christian Fri, 31 Jul 2009 13:12:17 +0000 http://actionable-analytics.com/?p=89#comment-54 Excellent point! However, your post then begs the question: How do you, with the aviailable data, identify the causes of conversion? Is it at all possible to do with web analytics? If not, what additional data would you want in order to determine the causes? Or, is the problem much more fundamental, namely that you are pointing to a limitation of statistics in general (hence your reference to the article from wikipedia? Excellent point! However, your post then begs the question: How do you, with the aviailable data, identify the causes of conversion? Is it at all possible to do with web analytics? If not, what additional data would you want in order to determine the causes? Or, is the problem much more fundamental, namely that you are pointing to a limitation of statistics in general (hence your reference to the article from wikipedia?

]]>