Prompting in onboarding to increase emergency savings

Background

Setting aside money for unanticipated expenses is a key component of financial health; it allows households to meet unexpected expenses without relying on expensive credit or fees.

Studies find that many Americans are lacking emergency savings; a 2019 study estimated that 53% of US households don’t have an emergency savings account. Qapital, a mobile banking app, allows its users to create personalized financial goals and save towards those goals using automated rules. Historically, emergency savings has only been around 4% of the Qapital goals for which users save, coming in behind travel, general savings, and debt, which account for more than half of the goals. During the COVID-19 pandemic in 2020, there was a near five-fold increase of new users with emergency savings goals, but there is still a large room for improvement.

We partnered with Qapital to see if we could encourage more new Qapital users to set up an Emergency Savings goal with a prompt at onboarding.

Key Insights

There were several ideas from behavioral science that we incorporated into our prompt design.

• Use defaults to signal norms. Research by McKenzie, Liersch and Finkelstein demonstrate that one mechanism in which defaults change behavior is by indicating to users the recommended course of action. We only let users skip setting up the emergency fund goal if they clicked “I already have an emergency fund.” The options create a norm that users should have Emergency Savings, whether in the app or elsewhere.

• Make it easy. Rather than asking users to create an emergency savings goal themselves, the prompt would create the goal for them with one click. Like participants in a study by Beshears and others on simplifying the retirement savings action to a single yes/no decision, this one-click option made it very fast and simple to set up the goal and start saving for it; users didn’t have to make choices on what to save or how.

We thought that these features in the prompt would encourage a lot more users to save for emergencies.

In addition to testing whether prompting users to create emergency savings goals was effective, we wanted to see the impact of different suggested target amounts.

Research on goals by Soman and Cheema suggests that the difficulty of goals impacts our likelihood of taking up the goal as well as our progress on the goal; difficult goals are seen as more important than easy goals, but goals must feel attainable for us to pursue. We hypothesized that users may be more motivated to pursue the emergency goal when prompted for the $100 target than the $500 target, since the $100 might feel more attainable

Experiment

We tested if a prompt to create an “emergency fund” goal in the Qapital app would increase the number of users with an emergency savings goal. New Qapital users were randomly assigned to one of three conditions:

1. Control condition: No prompt,

2. Treatment 1: A prompt condition with $100 target, and

3. Treatment 2: A prompt with a $500 target.

Users in the two prompt conditions would log into a flow with the emergency fund prompt, giving them an option to automatically create an “Emergency Fund” goal with a “Set and Forget Rule” that would transfer a defined amount into the goal each week. Both users in the control group and users that declined the initial offer to set up an emergency fund goal could later set up an emergency savings goal on their own.

Results

Over 6,500 Qapital users were part of the experiment to increase emergency savings goals with approximately 2,200 users per condition. Prompting users to create an emergency fund had a significant impact on the proportion of users that had an emergency savings goal with Qapital: the percentage of users with emergency savings goals in the prompt conditions were more than double the users with emergency savings goals in the control condition! There were no significant differences in the proportion of users with emergency savings goals between the $100 prompt and the $500 prompt conditions. The target amount may not have made a difference in take-up rate because the amount was not very prominent in the prompt or neither amount seemed particularly difficult to the users. In fact, in the control group, users who set emergency savings goals set targets of $3,000, $1,000, and $500. This suggests that the perceived difficulty of the target amounts in our treatment groups was likely not a factor.

In addition to having a higher take-up rate, we also found that users that had been prompted were more likely to actually contribute funds toward their goal. Within the 3-month observation period, only 18% of users with emergency savings goals in the Control group saved toward their goal versus 24% and 26% of the users in the $100 and $500 prompts, respectively. It’s possible that the prompt provided a norm that increased the importance of the goal for the user. There were not differences in the median amounts saved between the three groups.