Thursday, October 19, 2006

Google A/B tests for all

Google Website Optimizer appears to allow other websites to run A/B tests to optimize the effectiveness of their designs, features, algorithms, and content.

From their help page:
The Website Optimizer allows you to test changes in the website content of your pages in order to determine what will be most effective in getting conversions.

You choose what parts of a page you'd like to test -- headline, image, promo text -- and we'll run an experiment on a portion of your site traffic to determine which content on your site users respond to best.

When we've collected enough data, we'll provide you with reliable reports and a suggested course of action in order to optimize your site for maximum business results.
Much more detailed information in their technical overview.

It is no longer just Amazon and Google who have powerful A/B testing frameworks. Google appears to have just opened up their A/B testing tools to everyone.

[Found via Liam Morrison and Philipp Lenssen]


Anonymous said...

At least from what they show it isn't that advanced, you can pull that same bit of information out of your logs without requiring some javascript code with just a shell script.

Andrej Gregov said...

This is pretty cool. Although, all it looks like it just focuses on is clickthrough metrics. Decisions soley based on clickthrough can be dangerous. For example, say I test a new detail page display in my web store which hides the price of an item I'm selling. Clickthrough is sure to increase (for example clicking an Add to Cart button). But downstream sales may actually decrease because I've pissed off good customers who might not bother taking an extra click to see a price.

What is truley cool about this is that someone with relatively little programming knowledge and stats background can run controled experiments on their sites. Sure there are dangers in misinterpretting data. But metrics driven decision making is definitely a step in the right direction for (I'd guess) nearly all web sites which today make decisions based on intuition alone.