Smart use of A/B testing

Web Analytics and A/B Testing are amazing methods (inclined to see them as disciplines even) used for gathering data so that site owners or web analysts can look into it and come up with good ideas to improve the website. The Huffington Post decided to go one step further and use the extracted data from the A/B testing scenarios as part of the way the websites displays content. Now that is what I call actionable data.

Here is what Niemanlab.org reported 2 weeks ago about  The Huffington Post applying A/B testing to some of its headlines. Readers are randomly shown one of two headlines for the same story. After five minutes, which is enough time for such a high-traffic site, the version with the most clicks becomes the winner that everyone sees.

The concept could go even further if, lets say, they would apply that for segments, based on sources or if a visitor is new or old. The system would provide more than just one winning title and the end users would be satisfied cause the system seems to be presenting them just what they are expecting. OK, maybe I am overreacting a little as it would not be a piece of cake to develop all these, but lets just allow our imagination to run wild for a moment.

Maybe, in the near future

Google Website Optimizer the other week announced the release of their new API. At this point it can only create and modify different A/B or multivariate testing scenarios, but maybe at the next version it will be able to allow you to extract results as well. This way you would be able to replicate what The Huffington Post is doing in just a couple o hours and benefit from it on long term.

How willing would you be to use such methods for your own website?