Revealing Design Treasures From the Amazon - Web Design - Site Usability

Last week I had the privilege of going to An Event Apart in Boston where I participated in some great sessions on Web Design from some of the leaders in web design and development community.

The first session was from Jared Spool on Revealing Design Treasures from the Amazon. Jared starts his session by talking about Tuscan whole milk on Amazon, and the reviews included. Amazon stopped selling this milk a couple years ago, but people are STILL submitting reviews - and the reviews are HILARIOUS! A few examples Jared shared with us are reviews written from the perspective of a travel piece, a romance novel, and rock music review. The shorter ones are funny too, "OK product, but you have to buy a glass to use it".

"How many times have you gone into a planning meeting only to hear the words, "I like how Amazon does it, why can't we do it like them?" People don't say that about QVC... what makes Amazon so popular? A few stats:

A VERY successful site by any measuring stick. This isn't the only thing that makes Amazon so popular though - there are a few features that are nearly invisible that add a ton to the appeal of the website. An example of this would be from within the search results, the user can see when a product would be delivered. Most other sites do this on the product page, not the search results page. Pushing these results up higher allows the user to see important information without having to click deeper into the site. Another example of this idea is that Amazon provides incremental improvements rather than full, site-wide updates. Their shopping cart widget is a perfect example of this sort of iteration.

Based off these stats and the 'invisible' design iterations Amazon goes through, it's no wonder that people want to emulate them. But in order to emulate them we need to understand exactly what it is that Amazon does.

Engage Through Content

For Amazon, their content is their reviews. Shoppers will read the reviews that have been submitted on Amazon and then sometimes buy that product somewhere else. Amazon knows this and is okay with it. They even foster this by allowing people to continue adding reviews to their products that are no longer available on their website. Examples of reviews that are helpful not only to Amazon in driving customers to their sites, but the users that read them are two reviews from a Harry Potter book Jared shares with us. Once is about the content of the book while another talks about the experience this reviewer had with getting the book delivered from Amazon.

The problem with continually allowing people to add reviews is they are shown in a chronological fashion. This had disastrous consequences for Amazon because the crappy reviews were pushing down the good reviews. Naturally, Amazon saw this as a problem that needed to be rectified.

Their solution was a simple one. They added a simple question with a yes/no answer: "Was this review helpful to you?" That single addition to each review, is directly correlated to a $2.7 billion increase in revenue. This simple little addition didn't rollout with any big announcement of a new feature, but rather no fanfare. This addition allowed the more helpful reviews to filter to the top, while the crappy reviews fell to the bottom. Amazon found out that a simple chronological listing of reviews was not helpful to their users, but the helpful reviews were the most meaningful to their customers. Chronological reviews were only helpful to the author of the book and the people who wrote the reviews.

Some Experiments Don't Work

Although the key to success for Amazon has been experimentation, we must realize too that some experiments simply do not work. Amazon's interest in experimentation is extremely large because if an experiment works and is able to increase sales by $5 per order, that leads to an overall sales increase of $875 MILLION. If we as a web development community want to be like Amazon, we have to be willing to take incremental risks.

Amazon Gold Box

An example of an experiment that failed is the Amazon Gold Box at the top of the page that would wiggle at you. The idea of this 'widget' was to show that Amazon sells more than just books and DVDs. Essentially, they looked at your purchase history and show you things that are similar but nothing like you've purchased before and didn't know about. This frustrated people because they wanted to see things they were likely to buy. After awhile, people stopped using the Gold Box.


During the 'tag-craze' when every web developer who was worth their weight in gold started implementing tags, Amazon too jumped on the bandwagon. Some of the tags that people started using a lot were things like "book" and "DVD" but these are not very helpful, similar to the tag "not interested". Someone went through a lot of trouble to tag things that were not interesting to they could come back later and see all the things they aren't interested in? There are many other tags that people started using for bogus products, like "snakeoil", "IQ test", and "waste of money" are all tags that were placed on an ethernet cable that was selling for $500. Why would Amazon let their users tags things in such a non-helpful way?

The big lesson to be gleaned from Amazon's experimentation is this:

If your people are saying "we should be more like Amazon," that should really mean "don't feat new ideas." Experiment. Some experiments don't work. Know when to drop the ones that aren't working...

Jared leaves us with three points:

iAutoblog the premier autoblogger software