Logged-Out Recommendation

Two great articles on recommendation engines, here and here.

Helpfully, readwriteweb defines and classifies the different types of recommendation systems on the web thus:

• Personalised recommendation - recommending things based on the individual's past behavior
  
• Social recommendation - recommend things based on the past behavior of similar users
  
• Item recommendation - recommend things based on the thing itself
  
• A combination of the three approaches above

Amazon would seem to apply a combination of three approaches and is the site I want to focus on for the brief duration of this post

I use Amazon a bit and I just noticed that they don't let not being logged-in stop them from offering product recommendations. Great. I think that's great. If you use a  combination of PC's and a Mac (as I do) the obstacle of being logged-in to a website in order to see what it offers can be massive (even with Open-ID and all that stuff.)

So, how do they do it? Well, Amazon keep a brief history of your lat visit:

Then they make a mixture of social recommendations and product recommendations:

 So, is this useful? Ultimately, I think it is. Why is it useful? Well, I believe it is the difference between treating me as a random user and a customer. Even though I am not logged in I like that Amazon treats me as the latter. It's what I expect and frankly what I want to happen when I go and buy a product.