Bringing the User In Early | AutoTempest.com Design Case Study

Courtney Demone
Nov 23, 2017 · 5 min read

Role: Design Lead, Primary Prototyper, Lead UI Developer

For the responsive redesign of AutoTempest.com, we intended to release it quickly and iteratively. With our first release though, we discovered some of our choices weren’t quite what our users had in mind. This case study looks at ways to learn about your users’ needs in a way that compliments releasing quickly and iteratively. In the end, we saw:

  • 150% Increase in Conversions,
  • 50% Improvement in Mobile Bounce Rate, and
  • An Improved Iterative Research Process

AutoTempest aims to make the complicated and stressful task of buying a car more manageable, providing used car listings from all the major car classified sites in one place. With over 300,000 users, we were already an established favourite tool of car enthusiasts, with a #1 ranking to boot.


When the opportunity came to relieve some previous limitations of AutoTempest, we hopped at the chance to redesign it as a responsive web app. Our goals and intentions for the redesign were to:

  • Improve mobile accessibility by going responsive,
  • Release quickly, focusing on fine tuning and optimization in later iterations,
  • Optimize for conversions wherever possible, and
  • Pivot slightly to be more accessible to a savvy car buying audience.

Through quickly overviewing our user data, sketching ideas, exploring visual themes with style tiles, and iterating on in-browser prototypes, we created a product we were proud of and excited about. When we launched though, many of our users did not share our enthusiasm and our performance metrics floundered.

In our attempt to make AutoTempest more accessible to savvy car buyers, we had made a change so simple it didn’t even occur to us to have consider an alternative: we changed from keeping our results separated by source to sorting our results in one big list. This didn’t work for our users, as it made it more difficult for them to assess the quality of a listing and obscured our value. Once we had adjusted our new design to sort with separate results again, we realized there was a key thing that could have saved these frustrations: bringing the user into the design process earlier.

A very early in-browser prototype

Letting the user’s voice ring throughout the project.

With a set goal of releasing quickly and iterating on the redesign later, bringing users into the design process seemed a luxury we could live without for the initial release. We had done some guerilla usability testing and had provided tech support for years, and we were confident in our understanding of our users. That confidence had us conflate our assumptions with actual knowledge, which cost us on release.

By working with the user a little bit more, we could have challenged our assumptions and saved them a lot of frustration. This doesn’t necessarily mean the project needed thorough up-front research like we ended up doing that sort of research on SearchTempest. Through a few additions to our design process, we were able to bring the user’s voice into the design throughout the project, while still focusing on releasing quickly and iteratively.

  • Brainstorm creative solutions to even the simplest problems. When exploring the possibilities for a new design, we now start by challenging ourselves to think of creative ways to solve that problem, regardless of how unintuitive, technically infeasible, or plain ridiculous they may be. Thinking about your problem in divergent ways helps you challenge your current assumptions and leaves you with an array of jumping off points to accommodate for your users unanticipated needs.
  • I often divide my piece of grid paper into 9 sections, and give myself 15mins to fill all the sections with diverse designs. What would it look like if your search app could magically guess what the user was looking for? What would it look like if your blog was unable to scroll? What would it look like if your social media app made chat the top priority? These creative constraints can often lead to ‘aha’ moments.
  • Get your redesign ideas and prototypes in front of users early. A fast but thorough approach for this is to schedule a day at the end of each phase of your project to sit down with a couple of people and get their thoughts on what you have. to save time organizing, you can contact prospective interviewees all at once at the start of the project and schedule your interviews throughout. During your interviews, ask them questions about what your style tiles and copy make them think and feel, pass them print out wireframes and have them to navigate through them as if they were the real deal, and get them to click around in your prototypes before they’re fully functional. Be sure to provide your interviewees with compensation for their time! Premium access to your product or gift certificates often work great.
  • If you don’t have the resources for all that, you could get as simple as posting screenshots and ideas on your blog or social media with wireframes or style ideas and see how people respond. People aren’t always the best at knowing what will truly work for them, but it will at least give you a sense of whether you are in the right ballpark.
  • Gradually launch your new release. Start with an opt-in beta release and ask for people’s input. Then when you’re feeling confident enough to show to a general audience, slowly split test your release, starting with a small fraction of your audience, slowly increasing the audience size and making changes as needed. This way you’ll be able to get some qualitative feedback while also watching your analytic metrics for any quantitative drops.

The Results

On top of our improved processes, we achieved all our goals:

  • Conversions jumped by 150%,
  • Bounce rate on mobile devices dropped by 50%, and
  • Traffic to our suite of tools shot up 250%.

We were very happy with those results, and with a user satisfaction score of 89%, our users are too.