In 1935 Schrödinger stated that "If content on a website can be changed based a random factor, the website can simultaneously have many different states; yet to the observer there is only one state. Therefore, we throw in a bunch of observers, see what performs best and make some money". Schrödinger didn't take his theories any further (deciding instead to kill cats with cyanide) but Google did.
Google Website Optimiser (spelt with a z) is a tool for comparing variations in content, styles, layout and interactions on a webpage and testing what performs best.
Sounds great, but the problem is that it can be a bit of a pain in the posterior to setup with respect to the testing environment and complex scenarios. GWO can also be a bit daunting the first time, especially when a client is expecting a "vast improvement".
This session aims to:-
* explain the differences in A/B and multivariate testing, and when you can use them
* outline some the experiences I had setting up GWO
* show how to set up a test and review some of the complexities involved
* show some tips and tricks (and hacks?) for testing
* how to interpret the results
|