STC Usability SIG Home
Back to the Newsletter

This article was originally published in the May 2006 issue (Vol 12, No. 1)

About the Authors

Jacqueline Remus is Director of User Experience and leads the Shopzilla User Experience team in initiatives focused on creating passionately loyal customers.

Jessyca Frederick is User Experience Designer who champions the needs of the users by identifying creative new ideas that also support our business goals.

Shopzilla is a shopping search engine that aggregates and organizes more than 25 million products from more than 45,000 stores. See for details about Shopzilla.

STC UUX Community Newsletter

logo70x50.gif (1973 bytes)
Usability Interface

Million Dollar Web Usability Tips

By Jacqueline Remus and Jessyca Frederick

If you are like most User Experience (UEX) professionals, you struggle to justify to the rest of your company or design team what you know would be best for your users and your Web site. It can often seem as if the efforts of User Experience professionals are in direct competition with the goals of the business. What we have discovered here at Shopzilla, however, is that what has long been a struggle for UEX professionals can actually be a great tool to demonstrate the importance of your role. We have found a way, using tools that you may already have, to support the users’ needs that can positively impact your company’s bottom line.

Three key points for designing a successful user interface

1 - The business model. First and foremost, the goal of most businesses is to make money. By focusing on the way the company makes money, you can directly impact revenue. Presumably, your business model is already in line with the users’ needs, or you would not be in business at all!

2 - User needs. Focus on what your users actually want to accomplish on your Web site. It is important to keep the users’ goals separate from other competing goals, such as what you (or the business) want them to do on your Web site, or what your competitors are trying to accomplish on their Web sites.

3 - The relationship of metrics, user satisfaction, and good design. No matter how well an interface performs in qualitative studies such as usability tests or focus groups, the success of a Web site is ultimately determined by quantitative measures. A variety of metrics that may be appropriate for measuring success on your Web site, from business-oriented measures such as revenue and cost-per-click, to customer satisfaction-oriented metrics like page views, conversion and balk (or exit) rate. Successful user interface design can be measured by balancing the metrics that reflect both business goals and user needs.

Being mindful of each of these facets of UEX design will improve your site. Designing for all three will dramatically improve your results. Losing track of any of these three elements can often lead to mistakes that are costly both to the user experience and to the bottom line.

What this means at Shopzilla

Shopzilla is a comparison-shopping Web site. As a leading online marketplace, we have developed an iterative process that includes monthly usability testing, paired with a proprietary system that provides us with the type of metrics described above. We use those metrics to determine which user interface changes are “winners” (changes that will impress our users as well as our executives) and which changes are “losers” (those changes that will ultimately hurt the bottom line or the user experience). By balancing the qualitative data gathered in usability tests with quantitative data gathered through our proprietary system, we have fueled explosive growth at our company.

  • Our business model. It is simple: we are a referral service. We make money when a user “redirects”, that is, when they click on a link that takes them to the Web site of a store listed on Shopzilla.
  • Our user needs. Our users want to quickly find the products they are looking for at a reasonable price, while maintaining a sense of security in the world of online shopping.
  • The relationship of metrics, user satisfaction, and good design. We use our proprietary testing system to generate the metrics by which we judge our potential successes and failures. Our user success metrics are balk rate and conversion. Low balk rate is a good proxy for user satisfaction because if a user leaves our site without ever clicking through to a store listing, we can assume they did not find what they were looking for. If we can reduce the balk rate, we know that more people are finding what they are looking for.

However, a revenue increase (our primary business success measure) does not necessarily correspond to increased user satisfaction. If a user clicks on 15 store links but never buys any of those products (conversion), the experience generated revenue for us but was probably not satisfying to the user. Conversion, which occurs when users find exactly what they are looking for and purchases it through one of the stores listed on our Web site, is one of our best measures of customer satisfaction because we helped users buy the product they wanted.

At Shopzilla, user interface enhancements that are well-designed result in decreased balk rate and increased revenue and conversion. The following are three real-world scenarios that demonstrate the risks and rewards of adherence to these three guidelines.

Scenario 1: Introductory Copy

Problem: Many users come to our Web site after clicking on a link on Google or a similar search engine. As a result, they may not realize they are at Shopzilla and not at the search engine that they had just been looking at, or at an online store that sells products. Many users do not understand what comparison-shopping is or why it is important.

Proposed solution: Provide one to two sentences of introductory copy at the top of each page that explains where they are (the Web site and the category) and how comparison-shopping is going to make their online shopping experiences better. We tried several variations of text length and messages.

Usability Testing Results: In usability testing, it seemed the text was ignored altogether. Even worse, though, if we had run this introductory text on our Web site over the course of a year, it would have represented a loss of several million dollars in revenue and increased our balk rate. Clearly this idea was a loser to both the user and the business.

What happened? We did not meet the needs established by all three guidelines.

The business model. We forgot our business model! We depend on getting as many clickable, revenue-generating pieces of content above the fold as possible and we pushed these valuable UI elements further down the page by adding non-revenue generating content to the top of the page.

User needs. Find information quickly and easily. The introductory text pushed relevant content further below the fold making the page feel less relevant overall.

The relationship of metrics, user satisfaction, and good design. Our testing system told us that this text caused an increase in balk rate (that is, more people left our site without finding what they were looking for) and as a result our revenue went down. These quantitative findings support what we already know about good design—always keep the most important content above the fold.

Better Solution

Instead of pushing relevant content down by displaying a general message at the top of the page, we should integrate the message into the content that most interests our users.

Scenario 2: User Interface Tips

Problem: Users often report that they are “overwhelmed” by the information on our site’s crowded user interface. They frequently remark about the difficulty to determine which of all the possible actions will produce the results they want. Unfortunately, some of the most effective advanced features on our Web site go virtually unnoticed by users.

Proposed solution: Intersperse coaching tips throughout the page to tell the users how to get the most relevant content and identify the overlooked advanced features. We tried several different tip locations and styles.

Results: Had we run the user interface tips on our Web site over the course of an entire year, it would have cost us millions of dollars in lost revenue and increased our balk rate. This dismal result was confirmed by the fact that users did not notice the tips at all during usability testing.

So what happened? We did not meet the needs set out by all three guidelines.

The business model. Rather than focusing the users’ attention on our revenue generating content, we distracted them with clever comments about better ways to use our Web site. Our metrics showed us, again, that the best content on our site is that which actually earns money. That content is the reason users visit our Web site, and we should not do anything to pull their attention away from it.

User needs. Although we aimed to reduce their sense of being “overwhelmed” we actually aggravated the problem by adding yet another visual element to the page. We interrupted users while shopping for products on our site, thus decreasing the ease of attaining their goals.

The relationship of metrics, user satisfaction, and good design. The increased balk rate supports a basic tenet of good interface design: If the interface requires explanation—through a manual, tutorial, or even just screen tips—it is probably too complex and needs to be redesigned.

Better solution. Create new page layouts that are less overwhelming by reducing visual noise. Find a design that offers streamlined access to the most important content, reduces unnecessary information and advanced features, and focuses on our core capabilities: connecting shoppers and online stores.

Scenario 3: Number of Products Displayed per Search Results Page

Problem: Our Web site offers so many product listings that it requires the use of our advanced features to narrow down search results to retrieve a digestible set.

Proposed solution: Because the majority of our users do not take advantage of our helpful tools (e.g. filtering and sorting) and changing the number of products displayed on each page, we decided to change the appearance of how many search results were available by changing the default number of products displayed on a search results page.

Results: Reducing the default number of products displayed on search results pages was a huge winner for us! Happily, we see increased user satisfaction in our usability tests and our balk rate on these pages has decreased. Over the course of a year, we anticipate this small change will yield a several million dollar bump in revenue.

What happened? We followed our three guidelines!

The business model. Ultimately, our business goal is to connect buyers with sellers, to encourage our users to click through to the stores listed on our Web site. This beauty of this change was that it did not get in the way of any of our business goals, but still satisfied our users.

User needs. If users are overwhelmed by the number of results returned, they will balk instead of clicking through to the online stores we list. Finding a comfortable amount of information to present to users improved our click-through rates.

The relationship of metrics, user satisfaction, and good design . Once again, simplifying and streamlining the information on a page is not only good design, it is also good for the user and the bottom line. Ultimately, users want an interface to help them simplify their product search, but not obtrusive or overwhelming.


Anticipating the impact of changing the user interface is difficult, but balancing business goals, user needs, and proven design principals can increase your odds of making the right enhancements to your interface. It is crucial to validate your user interface changes with usability tests and metrics. All changes should have a quantitative proof of success to confirm that the changes are good for the user and good for the business. Best of all, demonstrating the measurable impact of your changes can help with championing usability testing and other user experience work in your organization!


All articles are property of the author or publication providing reprint permission. Reprinting this content in part or in whole requires permission from the source.
Go to STC Society Web Site