Abstract
<p>Friction-based digital interventions reduce unwanted smartphone behaviour by interrupting automatic app-opening sequences, but substantial attrition rates limit their real-world impact. One proposed strategy for improving retention is to gradually increase intervention intensity, allowing users to adapt before encountering the full intervention dose. We tested this in a preregistered six-week field experiment (N = 1,641) in which new users of the smartphone friction app one sec were randomly assigned to one of four delay escalation schedules ranging from immediate full-dose exposure (10 seconds throughout) to gradual escalation (3 → 6 → 10 seconds). Retention was low and relatively similar across conditions (65–70% dropout), with no significant difference between any of three escalation schedules and the standard full-dose condition. Exploratory analyses likewise found no association between delay duration and weekly dropout. However, among participants who remained engaged, longer delays were associated with higher rates of dismissed app opening attempts, suggesting that delay duration influenced behaviour without improving retention. These findings do not support gradual intensity escalation as a retention strategy and instead suggest that intervention characteristics may optimise different outcomes. While delay duration appears useful for shaping behaviour among retained users, retention itself may depend more on the introduction of interruption than its precise duration.</p>