A Simple Messaging Intervention Increases Completion of SNAP Applications

• In Event: Automation, AI and Nudges: Improving Digital Interfaces to Reduce Administrative Burdens in the Safety Net

Abstract

This study evaluates the impact of receiving a reminder message on the completion of Supplemental Nutrition Assistance Program (SNAP) applications. Building on a prior non-randomized pilot conducted with GetCalFresh (GCF)—California’s official digital benefits platform—we implemented a pre-registered, two-arm randomized controlled trial to test whether sending a reminder after application abandonment increases completion. Between December 11, 2024, and March 27, 2025, approximately 7,946 SNAP applicants who abandoned their initial application for at least 24 hours were randomly assigned to receive a reminder message either one day (treatment, n = 3,767) or five days (control, n = 3,879) after abandonment.

To estimate the causal effect of receiving a reminder, we focus on an early-only window—defined as the period after the treatment group has received a message but before the control group has. This approach isolates the impact of actually receiving a reminder, rather than simply being assigned to a timing condition. We estimate intention-to-treat effects on application completion and subsequent approval using differences in means and robust regression. Sending a reminder appears to have a 10 point effect on application completion (immediately prior to the control receiving the message the treatment completion mean was 28% with 95% CI: 25-32% and control completion rate was 18% with 95% CI: 16-20%). Results from the earlier non-randomized trial suggest that roughly half of these applications will be approved. These findings inform the design of a low-cost, easy-to-implement intervention to reduce administrative burdens and improve access to public benefits.

While recent evidence raises questions about the efficiacy of nudges, the results underline the value of the nature and value of nudges. In this case, a reminder from a trusted source (individuals had already provided information to the app), to complete a task that had already featured commitment of effort showed large effects. We argue that this is qualitatively different than, for example, texts to a broad potentially eligible population. A simple policy implications is that digital interfaces should provide users with timely reminders to complete basic tasks. 

Authors

• Jae Yeon Kim, Harvard University
  Presenting Author

• Eric Giannella, Georgetown University
  Non-Presenting Co-Author

• Zhaowen Guo, Georgetown University
  Non-Presenting Co-Author

• Pamela Herd, Georgetown University
  Non-Presenting Co-Author

• Sebastian Jilke, Georgetown University
  Non-Presenting Co-Author

• Donald Moynihan, University of Michigan
  Non-Presenting Co-Author