“When you assume, you make an ass out of u and me.”
The phrase may be a little sophomoric, but there’s a lot of truth there — especially that designs based on assumptions don’t just harm everyone involved: the designer, the stakeholders, and users.
Straying away from assumptions and into concrete, observable evidence is one of the pillars of growth-driven design. This somewhat new ideology favors a systematic, user-centered approach over clumsier method of saying, “hey, I want to include this trendy new feature because everyone else has it/I want one of my ideas in the design/my spouse said it was a good idea.”
For a complete review, read our article outlining the advantages of growth-driven design. Here we’ll just reiterate a few key points in case the concept is new to you:
The site launches quickly, and updates in small pieces with rapid iterations. This opposes the traditional model of launching after months of preparation and redesigning the entire site every few years. Updates and revisions are based on actual customer data, some of which are taken from the live site’s performance. The reality-based updates are more successful since they revolve around real customers, not theory or guesswork. Hence the name, “growth-driven design.” Growth-driven design puts more money in your pockets. Because the interface is hand-tailored to your specific shoppers, it streamlines and increases sales. Moreover, the frequent small updates spread out design payments over time, rather than periodic large lump-sums for complete site overhauls.
The whole of design, especially new features, should stem from facts about your user: what they want, and how they want it. To discern what they actually want from what you think they want, you need an empirical method to remove false leads. You need user testing.
Growth-driven design is only as effective as the quality of the data it draws from, and that puts a lot of emphasis on user testing. In this article, we discuss the five best user tests specifically for eCommerce.
We’ll start with the most basic — observing your customers use the site and seeing what they do naturally.
Direct user testing can be either moderated or unmoderated, remote or on-location. However, UX design and research consultant Chris Gray praises unmoderated remote user tests (URUT). URUTs offer a wealth of benefits over their counterparts:
Allows participants to test in their own environment, producing more natural results Participants can test any time they want, even late at night or in different time zones Participants can come from any location Loose scheduling requirements means less work for test organizers Results come in quickly Third-party testing resources can handle most of the heavy lifting, such as recruiting your target participants and analyzing the data
Regardless of how you conduct the test, giving the participant a specific task always helps focus the results. For example, you might request that testers “find the cheapest pair of ruby red slippers in a size 6.” Observing their behavior will shed light into how your typical shoppers navigate your site, compare prices, and use the interface.
Source: UserTesting via YouTube
Most testing services also offer video recording of the tests, with the participant’s movements on the screen, and sometimes even their facial expressions.
If you’re interested, third-party sites like UserTesting and Userlytics can conduct most of the work for you and deliver results fast. For more on preparing the test and choosing the right tools, read this Nielsen Norman Group article.
If you’re concerned about an individual topic or want to test the effectiveness of a single element, a more focused test like A/B testing works better.
A/B tests presents two different versions of a screen to different user groups (presenting two screens to the same user would askew the results) and marks the differences. This form of testing is ideal for layouts, visuals, and the placement of certain elements on the screen.
For example, is a call-to-action more effective in the top right corner or the bottom right? Which size attracts the right amount of attention? Which color?
Source: Optimizely via YouTube
With multiple elements at play simultaneously, eCommerce can benefit most from A/B testing. Optiverse released an up-to-date guide on A/B testing for eCommerce, covering all the areas you need to know about before starting the test, and even some opportunities you may not have yet considered.
For conducting the actual tests, try Optimizely for general screens, or Unbounce exclusively for landing pages.
Five seconds can tell you more than you think. In a five-second test, a participant is shown a sample screenshot (or logo, or any image, really) for five seconds. The image is then removed, and the participant is asked telling questions about what stuck with them.
These question can range from the general like “who is the company?” or test whether the participant noticed something specific, like “what percentage of discount is the sale?”
The reason the test is only five seconds is to see how well your visuals imprint on the mind of the shopper, even when the visuals are taken away. If your visuals communicate effectively, five seconds is more than enough time.
Source: ConversionXL via YouTube
As Craig Tomlin points out on his site Useful Usability, there are two main advantages of five-second tests:
They’re quick, easy, and cheap, plus multiple screens can be tested in each session. While you can hire a third-party to conduct them for you (see below), administering makeshift five-second tests on your computer to your colleagues and friends can be equally useful, as long as they fall into your target customer demographics. They provide both quantitative and qualitative data. Because emotional responses are more readily remembered, your participant’s “gut reactions” will stick with them once the image is removed. On top of that, you can ask precise questions to generate empirical data used for graphs, charts, or statistics.
The “original” Five Second Test site allows you to upload your images for testing, then conducts the test among its participant base and delivers a full report.
For eCommerce, information architecture carries a lot of weight. If products can’t be readily found in the assumed product categories, the shopper may become frustrated and leave for a site with more intuitive navigation.
Tree testing allows you to examine your existing site structure to determine if everything is where your shoppers want it to be. IA is highly variable, and different user groups think differently about product organization and categorization. Tests like these are the only way reveal the navigation preferences unique to your target customers.
Like direct user testing, you give the user a specific task and see how they solve it. However, because this tests narrows in on IA, they are only given text links to remove the distraction of visuals.
Source: Treejack via YouTube
If tree testing reveals a large amount of errors, you may need to rebuild your IA from scratch. In this case, try card sorting, the reverse of tree testing in which participants are given all types of products and asked to categorize them on their own based on their intuition.
Treejack is an online service that conducts tree tests for you. To learn more about preparing and conducting your own, read Martin Rosenmejer’s article on Webcredible.
From 2006-2009, Dr. Bob Bailey explored what he called the “firstclick” concept — the idea that a user’s very first click determines the outcome of their entire experience on a site. His results were as startling as they were practical:
If the user’s first click is correct, there is an 87% chance they will successfully complete their task. If the first click is not correct, the success rate drops below 50%.
Based on his findings, the first-click test was developed to improve the shopper’s success rate by improving their first click.
Source: Chalkmark via YouTube
The first-click test is easy to conduct and the data easy to read, although some services are equipped with the technology to go above and beyond. Chalkmark specializes exclusively in first-click testing, while Crazy Egg also monitors successive clicks and scrolling, and can present results in the form of a heat map.
The world of user testing is broad, with unique tests available to evaluate any conceivable area of your site design. However, in the realm of eCommerce, the five listed above have been proven to be the most useful. Don’t think of our list as a complete regime for user testing — think of it as a launching pad.
Do you have any questions, opinions, or advice on eCommerce user testing? Share your thoughts in the comments section now.