In addition to the material provided to us regarding Information Architecture, last semester, I came across this paper by Danaher, McKay, & Seeley (2005) which was about different types of Information Architecture (matrix, hybrid, hierarchical, and tunnel) and how they could guide behavior change. In the following paragraph, I would like to talk about four types of methods for determining what the right website is, and you can test different information architecture with these methods in order to scientifically determine which the best one is.
Let’s say that you are making a web-site that sells something and you want the user to buy it, so you want to modify their behavior to buy it, so the way you organize the website can influence on what they are going to click on, and influence when they click on the buy button. One thing the websites do is that they do analytics, A/B testing is one, and that’s where you have two versions of websites with different types of information architecture and different layouts, so they make and put both of them online and when you go to the website (let’s say that you want to sign up for credit cards), they randomly show you one of the different versions, maybe the A version or the B version (by versions I mean the type of Information Architecture design, for example: matrix, hybrid, hierarchical, and tunnel). And they see of the people who go to the version A, and the people who goes to version B, which one is more likely to sign up for the credit card, and they see which one works better and at the end that is the one that they keep.
Another thing that they can do it is what is called ‘multi-armed bandit’ which is like the more complex of A/B testing and multivariate testing. In A/B testing as I mentioned, one thing and two different versions are tested, but in multivariate testing you have lots of things that are different. Multi-armed bandit is more complex and mathematical way of doing that, so in that way you can see which version of the website brings about your intended behavior in the user which is the intended behavior is for them to sign up for the credit card.
There is also usability testing, which simply you show users your website, and you give them some tasks, for example to try to search for something or try to buy something or try to remove an item from the shopping card and then you take a video of their screen and you ask them questions like how hard was it, and you see how many mistakes they did and if they did a lot, then you know there is something wrong with your IA. This method is more anecdotal and is more expensive because you have to have users to watch.