It's often good to copy the big guys. When people grow accustomed to a certain design, following it in your own interface supports transfer of learning and thus increases usability. If you're designing an application, follow the lead of Microsoft Office — for example, use a floppy-disk icon to denote "save" (even though nobody saves to floppies any more). If you're implementing a search feature, copy Google.
For e-commerce usability, Amazon.com used to be the model. In 2001, we evaluated the usability of 20 e-commerce sites and Amazon was the clear winner, scoring 65% higher than the average of the other 19 sites. Having the Web's best usability served Amazon well: sales increased by 126% from 2001 to 2004.
Of course, rather than simply copy any one site, it's best to follow the hundreds of detailed research findings about e-commerce usability. But people often prefer to be told just one thing. For many years, that one thing in e-commerce design was "Do like Amazon." No more.
Amazon has recently changed so much that the average e-commerce site will reduce its usability by emulating its design too closely.
Paradoxically, Amazon's design may work well for Amazon itself. The company is simply so different from other e-commerce sites that what's good for Amazon is not good for normal sites.
What's Bad About Amazon
Cluttered pages. Amazon's product pages are littered with extraneous features, ranging from a "Gold Box" over a "wish list spree" to promotions for reading glasses and other irrelevant products. A single book page I analyzed contained 259 links and buttons. It was so cluttered that key product information — like publication date, page count, and average review rating — was three screensfulls below the fold (on a standard 1024 x 768 screen). Cluttered pages might work for Amazon because its users are typically long-time customers who know the features and can easily screen them out. Although first-time visitors are no doubt overwhelmed, by now they account for a tiny percentage of Amazon's revenues.
Internet-wide search feature. A small part of Amazon's clutter is caused by a search feature for the entire Web. This is a violation of guideline #52 for homepage usability: if people want to search the Web, they'll go to their favorite search engine, not your site. Amazon nonetheless offers this feature because it owns the featured search engine, A9. Promoting A9 likely provides sufficient business benefits to the overall company to compensate for lost sales caused by clutter on the main site. But if you don't own a search engine, don't offer a Web-wide search.
Advertising on product pages. Amazon spends about two inches of each product page advertising other websites. Although this generates revenue, the average e-commerce site should be ashamed if it can't make far more money selling to a hot lead who's already investigating one of its own products. Amazon's position as the default place to buy books is so strong that it can afford to send shoppers off to other sites, knowing they'll return later and buy the book anyway. You can't make the same assumption. Sell to your prospects, rather than throw them away.
Lousy UI for specialized product categories. Try buying a Mozart concerto or a plasma TV on Amazon. Unless you know in advance the exact product you want, Amazon's category pages make it basically impossible to identify the best offerings. Using a single e-commerce engine for thirty-one different product categories has fueled Amazon's tremendous growth. The cost to customers, however, is poor support for each product type's special requirements. If you're selling classical music, you're better off studying the best classical sites rather than copying Amazon's design. Something that works for popular music fails for a genre in which a single composer can be represented on thousands of CDs.
Lack of integration with its international sites. It's good that Amazon offers fully localized sites in six countries outside the United States, but there is no integration across the sites. If you're a French user looking at a book on the American site, for example, there's no indication that you can order the same book from Amazon.fr or that a French translation is available. Also, third-party sites that recommend books are relegated to extremely awkward linking strategies when they want to support international users (for example, "Get this book from Amazon in the U.S., in the U.K., or in Japan").
Co-branding. Amazon is leveraging its e-commerce engine to serve as the online platform for many other companies. Much confusion results, as users don't know what company they're dealing with. People understand the simple model of one company, one website. Anything else is problematic. For example, sometimes you think you've ordered from Amazon, only to get email from a company you've never heard of. In these times of overflowing inboxes, we know that users are likely to delete such emails unread. Usability problems also result when users are interested in a company that outsources its e-commerce to Amazon. For example, we tested Toys-R-Us as part of our "clicks-and-mortar" study of store finders and locators: because Amazon handled the company's online toy sales, people had great difficulty finding the address of a real-world toy store on the website.
What's Still Good About Amazon
Confirmation email. Except when they're part of a co-branded sale, Amazon's use of confirmation emails keeps users in the loop and increases their trust in the overall fulfillment process.
Fulfillment. When you buy something, you get it. In the rare case that there is a problem, you get a clear email telling you so. Of course, that's the way e-commerce should be, but all too often it's not when you shop on other sites. Amazon's success with fulfillment clearly proves that the total user experience goes beyond the user interface.
Login screen. Amazon's sign-in screen remains a model to be emulated, minimizing the common problem of new customers who try to log in without having registered. Amazon presents two questions in linear order: (1) "What is your email address?" and (2) "Do you have an Amazon.com password?" For the second question, users can select one of two radio buttons: "No, I am a new customer," or "Yes, I have a password." Many other sites present the new- and established-user sections side-by-side, and thereby divert new users to the established-user section through the magnetic attraction of type-in fields.
Relevant cross-sales. "Customers who bought this book also bought" continues to be a stellar approach to cross-selling. It almost always drags up other relevant products that the customer is likely to want.
Sample content. One of the biggest challenges in e-commerce is alleviating people's fear of getting either the wrong product or a low-quality one. Online sales have a huge handicap here, because shoppers can't touch the merchandise. To minimize this problem, get users as close to your products as possible in a virtual world. Give them the specifics they need in a language they understand and show them meaningful pictures. Amazon does this well, especially in using sample content, such as book excerpts and music snippets.
Comprehensive product selection. Amazon lays claim to "earth's biggest selection" and for once, a company actually lives up to its slogan. We've known since 1997 that users want comprehensive services on the Web, and Amazon fulfills that need, especially within their original focus of books. If it's published, they carry it. Even better, they don't remove a book from the site once it goes out of print; instead, they offer a marketplace for buying and selling used copies. The product page retains the same URL, even when the product status changes, thus avoiding the linkrot that plagues so many other etailers.
On balance, Amazon is still the world's best e-commerce site. Many of its strengths, however, are unique to its status and would not carry over to sites that emulate its design.
The full report series with all our usability research on online shopping with actionable design guidelines for e-commerce sites is available for download.