We have done what I think is some pretty extensive A/B testing using wicket.  
As someone suggested, you can use Google Website Optimizer/Google Analytics to 
do some A/B testing.  I found that approach is great for simple testing such as 
Sentence A vs Sentence B or Color A vs Color B.  If you need more advanced A/B 
testing such as page flow, etc.., the google platform will not be as easy to 
use.  The localization-and-skinning of pages is also a possibility but that 
will only get you look and feel changes.

We performed A/B testing on a page flow [workflow] and large content blocks.  
This was all handled by wicket and we used google analytics to track the events 
and page views.  When a session is created, we assigned the user randomly to an 
A or B group and set a cookie so that on future visits they would be in the 
same group.  There is a bit more detail in this as we also assigned a test 
number so that when a new test was run on a different area of the site we could 
re-assign the groups.  Then in a base page we pushed some javascript out 
basically setting the test number and A/B group out to Google Analytics using 
custom variables.  In the area of code that we wanted to test we would have an 
if A then do A group stuff else do the B group stuff.  This was usually on a 
link or click event but we could also show different panels this way or set 
large blocks of WebMarkupContainers to not visible using this approach.  Google 
Analytics events were fired as needed to track events or flow between pages.  
Once the framework was in place we could quickly run an A/B test and turn it 
off just as quickly with a flag to enable or disable A/B testing.

One thing to keep in mind with A/B testing is that you need to test one 
discrete thing at a time otherwise you begin to do multi-variant testing which 
is possible just much more complicated.  In either case brush up on your 
statistics knowledge to analyze the data and know when you have enough 
statistically relevant data or you will be making decisions on what could be 
bogus data.



-----Original Message-----
From: Fergal Keating [mailto:fergal.keat...@directski.com] 
Sent: Friday, April 20, 2012 6:10 AM
To: users@wicket.apache.org
Subject: Re: A/B testing with wicket

You can use style and variation. Setting default style works automatically to 
load differing HTML files for the same components.

https://cwiki.apache.org/WICKET/localization-and-skinning-of-applications.html


On 20 April 2012 09:08, Decebal Suiu <decebal.s...@asf.ro> wrote:

> Maybe Google Website Optimizer/ Google Analytics and not Google AddWords.
> My
> idea is to use Component.getVariation() as start point.
>
> Best reagrds,
> Decebal
>
> --
> View this message in context:
> http://apache-wicket.1842946.n4.nabble.com/A-B-testing-with-wicket-tp4
> 571946p4573185.html Sent from the Users forum mailing list archive at 
> Nabble.com.
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscr...@wicket.apache.org
> For additional commands, e-mail: users-h...@wicket.apache.org
>
>


--
Fergal Keating
IT Senior Engineer
-----------------------------------------------
e. fergal.keat...@directski.com
p. NA
w. www.directski.com

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@wicket.apache.org
For additional commands, e-mail: users-h...@wicket.apache.org

Reply via email to