Thursday 21 May 2015

How We Redesigned Yelp’s UX With Remote Usability Testing

Usability testing with real users, is one of the key areas that many web designers skip. Often, there is little money in the budget for testing, but usability testing distance can be relatively inexpensive to implement, and the results can make a big difference to a site conversions.

To demonstrate the power of usability testing in the design, we partner with User Testing and optimal Workshop to run some tests on Yelp as part of a redesign exercise.

Specifically we chose remote usability testing because they are quite fast and quite affordable to run, compared with focus groups and other testing laboratory based design. These tests were all unmoderated so users can interact with the Yelp website in the comfort of your own home for the most natural results. Encouraged to think aloud and then recorded their reactions.


User research is not writing complex reports - it is asking the right questions and use evidence to support design decisions. In this piece, we'll see how we elect our users, how we set the tasks and what were the results ...

Selecting Users:

One of the first steps is to find out who is testing their designs. In our experience, we have found that demographics are not as important as the behavior and familiarity with technology. How often they do use similar platforms, and how comfortable they are?

Yelp has a base of large users (138 million unique monthly visitors, according to the Q2 of Yelp 2,014 numbers) to our redesigned site still it had to be familiar to the average current user - no sense to alienate power users for there to woo first-time users.

We do not focus on age, sex, income level, or experience in the use of the Web, because Yelp users come from all backgrounds. Because we are handling qualitative data, no need to worry about statistical significance. We follow the best practices in the industry and we ran our study with a total of 5 users who help reveal about 85% of the usability problems (good enough for exercise).

One task required users to log into an account. This meant we had to create two segments of our test base: one with accounts Yelp (3 users) and one without (2 users). For Yelp segment accounts only selected participants who were users Yelp for less than 6 months to clear even more likely to be power users.

Finally, for the sake of simplicity, we only tried the website Coffee in desktop, not mobile devices. (If this was more than just an exercise, we would have tried the experience on as many devices as possible.)


Creating tasks for our users

Each usability test should start with the question: "What do we learn?"

For us, we wanted to learn Yelp semi-frequent users complete common tasks (to identify what features were most important) task and at least one less common (to test the intuitiveness of advanced features).

We gave all users of such common tasks:

  •     Focused task - Find a very specific parameters based on business
  •     Open-ended task - Find a business without giving many guidelines
  •     Very specific task - Find a specific place to learn a specific piece of information

We wanted to learn when both groups of users chose to look in front of sail, how they interacted with filters, and how the decision about which business to visit was taken.

As for less common tasks, we provide a different task for each user group.

As we had heard several complaints from users about Yelp registered Bookmark and lists features, we ask the registered users (Group 1) to save companies for later reference. For unregistered users (Group 2), we ask them to find an event.

Then all the tasks we have assigned to both groups are presented. After each test, we ask users if they were able to complete the task successfully and the level of ease or difficulty (known as the question Ease Individual).

    Imagine you need to book a private dining room for a group of 15 people. Are you looking for an Italian restaurant with an elegant atmosphere? Its budget is about $ 20 per person. Try to find a restaurant near you that matches all these requirements.

    Imagine you are driving through Boise, Idaho, and your car starts making a strange sound right when you're about to stop for the night. His companion recommended 27th St Automotive. Use Yelp to find out if they are open at 8:00 pm Tuesday.

Task Group 1 (account holders Yelp)

    Imagine that your best friend is having a birthday soon, and you be planning a party. Find 10 bars or lounges near where you live would be interesting to study later for the party. Save them so that you can easily find it again on Yelp.

    Go to the place where you saved the 10 bars for the party of his best friend. Keep your tastes in mind, choose one that would be a good match.

Task Group 2 (non-account holders)

    Use Yelp to find a new restaurant near you that you have not been to yet. Do not spend more than five minutes looking for.

    Imagine you are looking for something fun and unique to do in your neighborhood this weekend. Try to find a concert, play, or other event using Yelp.

Once this was done, it was time to begin the Pass4sure SY0-401 test. It took about an hour to get the results before we could see the reactions of users and analyze data.


The breakdown of data usability

To compare qualitative data we had now, we have a quantitative test with 35 users with a closed-type card and proof of first click. You can learn more about quantitative user tasks, but we'll just summarize the main points of view:

    The search bar was the starting point for almost all tasks. It was also the option preferred backup when users were unsure of how to interact with the site user interface (for example, search for "Bars" instead of clicking on the category). Our new design definitely need to prioritize the search bar.

    The Events tab not noticeable. When asked to find an interesting activity, a user was to search bar, while the other sailed through the top section of Yelp. If we wanted users to interact effectively with the events told in Yelp, we need to make it easier to find.

    Price categories were unclear. When a budget is administered find a restaurant, some Users were not sure what they meant dollar signs. In our new design, we have added price ranges to symbols.

    Filters are not prioritized correctly. People are not used 7 Coffee filters 47, and the most popular filters that emerged in the tests (eg, "accept credit cards" and "Open Now") take multiple clicks to access. Our redesign reorganized in groups of 4 filters for easier access.

    The photos are a key part of the experience. When asked to find restaurants with a certain environment, users relied on photos. Our new design makes more visual Yelp.

    Markers need be simpler. Currently, you cannot just keep a restaurant or business directly from the search results - you need to visit each individual page marker. Our new design allows you to save a company with a single click on the search results page.

3 comments: