Thursday, February 3, 2011

Amazon’s personalization feature should be more personal

Amazon is an ecommerce trailblazer. From the long tail strategy that Jeff Bezos adopted for the company to its groundbreaking affiliate scheme. Where this online giant leads, others follow, including Packt.

Amazon was one of the first websites to introduce personalization and the ability to push targeted products to customers based on their browsing and purchase history. This transformed the customer experience and gave us products and options that we weren’t aware of and ultimately persuaded us to buy more.

I like French films but my French isn’t great, so I rely on reviews and recommendations from English speakers. With Amazon’s personalization, I get a regular stream of suggestions and ideas of films to watch. However I think Amazon’s personalization algorithm is flawed and needs updating.

My main gripe is that it doesn’t take into account fast-moving, regular purchases against one-off or once-in-a-while items. Before expanding into the world's biggest department store, Amazon's core business was books, CDs and DVDs. Personalised recommendations of these reasonably low-priced items work well and I have bought many a French film thanks to them. However if I bought a new LCD TV, I don’t expect to be presented with a list of other LCD TVs every time I visit the site over the next few weeks.

Amazon also heavily weights its personalization results with your most recent activity. So if I was to visit any page on its website, even if someone sent me a link to a product with funny reviews, my next session would be dominated by similar products. One of the reasons I included this here was because I had been sent a link to the reviews of an album by a questionable celebrity duo and for my next visit, I was presented with the male half of the duo's back-catalogue. This has happened with other products.

However on researching this blog, I found a link to an article about funny Amazon reviews. I checked out this relaxation tank on their .com site, with some very funny customer comments. I then visited the homepage to see what I was presented with, expecting other relaxation-based products and instead, there was a list of the other items with funny reviews. Maybe Amazon has cottoned on, with its .com site leading the way.

Over Christmas, I used Amazon to buy many of my presents, as many of us did, however even though I was happy to buy my sister a copy of the Style Book: Fashionable Inspirations, I didn’t want to see similar products the next time I visited. I want to see suggestions of similar products that I'm interested in, not those that I've bought for others. My sister may disagree.

With Amazon, you can delete items from your browsing history, however for an intelligent system, I shouldn't have go out of my way to visit my browsing history and then scroll down to delete the things I wasn't interested in seeing. They could make this easier for customers by offering checkboxes on product pages and in the checkout procedure. Even if there was an option for me to state whether the purchase is a present and if I wanted to remove it from my browsing history, would be a good start.

Amazon, with all its customer data and purchase history, should also understand what products are bought regularly and those that we buy less frequently. These products should be removed from the personalization adverts, or at the very least, show me one or two products, not a never ending list. For what has been designed as an intelligent system, it hasn't scaled its intelligence as the site has scaled.