Notes from 56th International STC Conference
Atlanta, Georgia, May 2-5, 2009

Revealing Design Treasures from the Amazon

Jared Spool, User Interface Engineering

Session Description: On the surface, Amazon.com just seems like a large e-commerce site, albeit a successful one. Its design isn’t flashy, nor is it much to write home about. But deep within its pages are hidden secrets—secrets that every designer should know about. In this entertaining presentation, Jared will share some of UJIE’s latest research into the hidden treasures of (the) Amazon.

• Introduction
   
• Amazon.com has become a product review forum even for many products they do not sell
• Initial example was Tuscan Whole Milk (all kinds of reviews, parodying many literary styles)
• Amazon has 8th highest traffic on the Internet; $19B in sales in 2008, up 30% from 2007 in terrible economy
• Amazon has third highest Foresee customer satisfaction rating.
• Amazon has 76M customers
• Processed 175M orders last year, 24 per second
• Searching on Amazon: e.g., cameras, pulls up product photo, price, commitment on arrival time (several options on speed of delivery)
• Shopping cart has been redesigned several times to maximize revenue
• Four treasures about Amazon follow
• Treasure 1: Engaging through contact
   
• Harry Potter series has thousands of reviews per title
• Traditionally, reviews have been placed chronologically, pushing the best reviews off the screen.
• Amazon adjusted reviews in a way that raised $2.7B per year. Added a question “Was this review helpful to you?” Despite a 1-in-500 response rate, the best reviews were still identified and pushed to the top — greatly boosting sales. Customers often focus on negative reviews to assuage “pre-buyer’s remorse”
• Caution: emulating features of Amazon doesn’t necessarily work. Target “bought” the entire software application (the shell). All Amazon had was negative reviews. Problem was people assumed most of Target’s product line would work; satisfied users weren’t motivated to write a review, but dissatisfied users did
• Target wasn’t nearly as successful at reviews.
• It takes 1,300 purchasers to generate 1 review.
• Treasure 2: Don’t fear new ideas.
   
• Amazon constantly experiments with new ideas; not all of them work (like the blinking Gold Box in the upper right-hand corner) ... users could click through 10 items within an hour and get a special price, but they couldn’t back up among the 10 and therefore couldn’t compare. Also, they didn’t target the market; they were pushing products other than books, CDs, etc.
• User review tags can be very revealing, as when they tried to sell an Ethernet cable for $500.
• Moral: you have to be willing to fail in order to find the big successes (managed risk)
• “Defective by design” was a major tag
• Treasure 3: Eliminate tool time while delivering confidence.
   
• Amazon presents options before asking for password; even sales. This can be done with a very sophisticated security system. Level 0: doesn’t know who you are (no cookie); will ask for password; Level 1: They know you as past customer; “if you’re not Jerry, click here.” Level 2: Amazon checks out things that only you would know (change in address)
• Goal time ...when the user is improving the outcome of the experience
• Tool time ... when the user is moving forward without any improvement in the outcome of the experience. Very friendly in remembering passwords ... user-friendly.
• Changes must be subtle or customers will not like them.
• Amazon 2007 navigation phase-in plan: Phase I: tested with 5,000 non-cookie users; Phase II, 1 of 5 non-cookied visitors; Phase III, Cookied customers, 5,000 per day; Phase IV; 20% of cookied customers; Phase V: open to all
• Amazon doesn’t fear experimentation but is very cautious in user-testing it.
• Caution: Not every use case is the same. Some items are easy to find with a search (very specific). Can be harder: The first Tom Clancy book featuring Jack Ryan; an inexpensive but high-quality SLR camera; a good toy for my 6-year-old niece; novels written by Nobel Prize for Literature winner
• CD Baby: search result will be accompanied by an editorial viewpoint
• Treasure 4: Never forget the business
   
• Amazon uses “loss leaders” to attract users. It can turn a profit even selling its entire inventory at cost due to very rapid turnaround on sale.
• “Negative operating cycle” ... Amazon turns its inventory every 20 days (Best Buy, 74). Standard retail payment terms are 45 days. The lost time puts Best Buy in the hole; the gained time brings Amazon a substantial amount of interest based on the size of their revenue flow.
• Playmobil security checkpoint: check key words and reviews
• Final thoughts
   
• Engage your users by delivering great content
• Don’t fear trying out new ideas (“a risk-averse company sells crap”)
• Eliminate tool time while delivering while confidence
• Never forget the business
• Caution is warranted
• Be careful when emulating
• Some experiments don’t pan out
• Not all use cases are the same