Policy & Practice December 2018

Chart 2 Survey among a random subset of Fresh EBT users, October 2018

design practice—the continual process of discovering and addressing user

In addition to online search, Facebook has proven to be an effective way to reach clients. Among Fresh EBT users, Facebook is the most popular mobile app and social network. More than three-quarters of Fresh EBT users report using Facebook at least weekly, 63 percent say it is their most used social media platform (see Chart 2), and nearly half say that Facebook Messenger is the most common way they chat with friends and family (even more so than regular text message). Lesson Learned: Social media plat- forms can be one of the most effective tools to reach your clients, but not all social media platforms are created equal. Prioritize Mobile Websites While Fresh EBT is a cross-platform service—it works on both iPhones and Androids—more than three-quarters of users have Android phones. 4 On average, Android devices tend not only to be less expensive, but also have less storage space, are slower and are usually on more limited data connec- tions 5 than iPhones. Fresh EBT users frequently run out of space and data on their phones and aggressively delete apps that take up too much space or use too much data. Even more so, the Android operating system specifically recommends that users delete big apps and download small ones to save space on the device. Thus, it is exceptionally important to consider prioritizing small size and speed over advanced functionality. One way to avoid this challenge altogether is to build a mobile website instead of a mobile app so users don’t have to download or store anything on their phones at all. Despite seemingly similar functionality, mobile websites and mobile apps actually fill very dif- ferent roles’ in people’s lives. The typical American uses a fewmobile apps a lot— more than two hours per day in their top five apps—and a lot of websites a little. 6 Most mobile apps are either used weekly or uninstalled altogether, and nearly 80 percent of all mobile apps are uninstalled within three days. A common best practice is to build a native smartphone app for services that people use daily or weekly and build a Specifically for Low- End Android Devices

needs—for the past two years. An early survey revealed that a

majority of Fresh EBT users were actively looking for job opportunities. We learned that most Fresh EBT users were single moms who needed flexible hours. To address this need, we built a simple job board inside Fresh EBT that highlighted local jobs with fair and flexible hours. Initially, we saw high user interest as measured through click-through-rates, but relatively low application rates. People were viewing jobs but not applying to them. We began helping users through the process and discovered a perhaps unsur- prising barrier. While the job listings were smartphone-friendly, the job applications were not. Considering low- income Americans are four times more likely than the average smartphone owner to apply for jobs on their smart- phone, 7 the design of employer websites represented a significant barrier. Employers with mobile-friendly applica- tion websites had application rates 20 times higher than those without. In the latest iteration, we have significantly pri- oritized job listings for employers with mobile-friendly applications. With this change we saw a significant increase in the number of job applica- tions completed, but lower engagement with the job board than expected. Consequently, we ran a few experi- ments in which some Fresh EBT users saw the standard user interface and others saw a slightly tweaked version. Tweaks included altering the types of jobs offered, changing the job descrip- tions, and inviting clients to narrow job listings by specific criteria earlier in the process. Finally, we shared information about income limits and monthly SNAP allotments to show how an increase in income does or does not affect SNAP eligibility (see Chart 3, page 35). The users who saw the SNAP and income information were more likely to search or tap on a job posting and more likely to return to that posting to apply. Though this was a small test and we have more to learn, our hypoth- esis is that providing this information helped our users feel more confident

mobile website for services that people use less than weekly. Considering the frequency of many common transac- tions in human services—annual recertifications, biannual renewals, quarterly reports—a small mobile website optimized for low-end Android phones is often the most cost-effective and client-centered approach. Lesson Learned: As much as possible, optimize smartphone applications to work best on Android phones—make them small and fast. And, for services used less than weekly, consider mobile websites instead. Continually Discover and Address Evolving User Needs Over the past three years, we have updated Fresh EBT every month. Each update has included a combination of small iterative improvements and new feature sets based directly on user feedback. To make this process concrete and actionable, this section reviews the evolution of our efforts to help Fresh EBT users earn extra income. Fresh EBT started exclusively as an EBT management app. It helped card- holders check their benefit balance and monitor their transaction history without calling the EBT customer service phone line. Upon this foundation we then added a simple survey platform to begin soliciting feedback and suggestions directly from clients. This informa- tion has fueled our client-centered

See Downloads on page 35

December 2018 Policy&Practice 27