Repairable Streaks, Not Strict Streaks or No Streak: Missed-Day Handling in AI-Coached Daily Practice and Who Comes Back After a Break

Abstract

We randomly assigned 402 adults starting a 12-week AI-coached language course to one of three streak designs and followed their logged practice to week 24. In "strict-streak" accounts (n=134), the coach counted consecutive practice days and reset the count to zero after a missed day. In "repairable-streak" accounts (n=136), one missed day a week was absorbed, and a further missed day could be repaired with a five-minute catch-up session within two days. "No-streak" accounts (n=132) used the same coach and lessons with no counter. Strict-streak participants practiced most in the first four weeks (5.6 sessions a week, against 5.2 for repairable-streak and 4.4 for no-streak participants), but over the full 12 weeks repairable-streak participants completed a mean 46.9 sessions, against 44.6 for strict-streak and 39.8 for no-streak participants. By week 12, most participants in every group had a break of two or more days; 59% of repairable-streak participants who had one returned within seven days, against 46% of strict-streak participants (a 28% gap) and 52% of no-streak participants. In weeks 21 to 24, 40% of repairable-streak participants were still practicing every week, against 31% of strict-streak participants (a 27% gap) and 33% of no-streak participants (a 19% gap). Participants who had missed a day by week 6 rated how badly they felt about it at 4.1/7 under the strict streak, 3.2/7 under the repairable streak (strict-streak ratings 28% higher), and 2.9/7 with no streak.

1. Introduction

Of the game elements an AI coach can add to daily practice, the streak counter is the simplest and probably the most widely copied: a number that grows by one for each consecutive day of practice. Its appeal to the people who run practice programs is easy to see in the first weeks, when a growing number gives learners a reason to open the app on a tired evening. Its weakness shows later. A streak that resets to zero after one missed day turns an ordinary interruption, such as an illness, a late shift, or a child's school play, into the loss of everything the counter had recorded, and a learner who has just lost forty days may find it easier to stop than to begin again at one.

Repairable streaks keep the counter but change what a missed day costs. We compared a strict reset, a repairable design, and no counter at all, with the question of who returns after a break as the main concern.

2. Method (Summary)

Participants were 402 adults enrolling in a free 12-week language course run by a municipal adult-education service, in which an AI coach set daily lessons of 10 to 20 minutes, gave spoken and written feedback, and suggested review material. At enrollment, participants were randomly assigned to one of three account designs. In strict-streak accounts (n=134), a counter on the home screen showed consecutive days with at least one completed session; after a day without one, the counter returned to zero and the coach's next message noted the reset. In repairable-streak accounts (n=136), the first missed day in any calendar week did not break the streak. A second missed day in the same week could be repaired by a catch-up session of at least five minutes within the following two days, and the coach's message after a miss offered the catch-up without mentioning a reset. A streak broke only if the catch-up was not done. No-streak accounts (n=132) had the same coach, lessons, and reminders, with no counter.

Practice was taken from the coach's session logs, with a session counted when a lesson of at least five minutes was completed. Catch-up sessions met that threshold and were counted as sessions for all practice and return measures. A break was defined as two or more consecutive days without a session, and a return as any completed session within seven days of the break's first day, assessed for each participant's first break. Participants were counted as still practicing in weeks 21 to 24 if they completed at least one session in each of those four weeks; the course's lessons remained available after week 12 with the account design unchanged. At week 6, participants who had missed at least one day by then (97 strict-streak, 101 repairable-streak, and 99 no-streak participants) rated, in an online survey, how badly they had felt after their most recent missed day (1 = not at all, 7 = very badly).

3. Findings

The strict streak bought a faster start. In weeks 1 to 4, strict-streak participants completed a mean 5.6 sessions a week, against 5.2 for repairable-streak participants and 4.4 for no-streak participants. Over all 12 weeks of the course, the order changed: repairable-streak participants completed a mean 46.9 sessions, strict-streak participants 44.6, and no-streak participants 39.8.

Breaks were common in every group. By week 12, 112 of 134 strict-streak participants, 108 of 136 repairable-streak participants, and 111 of 132 no-streak participants had at least one break of two or more days. Of those, 64 of 108 repairable-streak participants (59%) returned within seven days, against 52 of 112 strict-streak participants (46%, a 28% gap) and 58 of 111 no-streak participants (52%).

In weeks 21 to 24, 54 of 136 repairable-streak participants (40%) completed at least one session every week, against 42 of 134 strict-streak participants (31%, a 27% gap) and 44 of 132 no-streak participants (33%, a 19% gap). Strict-streak and no-streak participants were close to each other.

Among participants who had missed a day by week 6, ratings of how badly they felt afterward averaged 4.1/7 in the strict-streak group, 3.2/7 in the repairable-streak group (strict-streak ratings 28% higher), and 2.9/7 in the no-streak group.

Repairable-streak participants did not treat the allowance as a loophole for skipping practice, but some did settle into it: 31 of 136 (23%) used the absorbed day in at least eight of the twelve weeks, most often practicing six days rather than seven in those weeks. Catch-up sessions followed 63% of the second missed days that the repair rule covered.

4. Discussion

The number we would put first is 46%: fewer than half of the strict-streak participants who had a break came back within a week. Under a strict reset, every one of them had just watched the counter return to zero, and the coach had said so. The repairable streak and no counter at all produced higher return rates, and only the repairable streak also kept the early lift in practice that made the counter attractive in the first place.

That combination is what we would expect if the counter's value lies in the record of effort it holds and its harm lies in what happens to that record after a lapse. A counter that survives an ordinary bad week still gives the learner something to protect, and one that resets tells a person who missed a day that the forty before it no longer count. Neither streak design asks for perfect attendance in words, but only the strict streak treated a missed day as a failure.

The repairable streak had a cost, and it fell on the program. Its weekly practice figure in the first month, which the adult-education service reported as its main engagement measure, was lower than under the strict streak (the strict streak's 8% higher), and nearly a quarter of its participants regularly practiced six days rather than seven. The service would have had a better-looking first month with the strict streak and fewer of its learners still practicing by week 24.

5. Limitations and Future Directions

The 31 participants who used the absorbed day in most weeks may differ from those who did not in ways we did not measure, so we cannot say whether the allowance created the six-day pattern or let people already inclined to it keep practicing. Because catch-up sessions counted as sessions, the repair prompt itself encouraged an early return, so the return measure partly reflects the design rather than independent re-engagement. The repairable design also combined an absorbed day and a repair option, and we cannot say which of the two produced the higher return rate. The return measure counts any session within seven days and does not distinguish a single session from a lasting return. Ratings of how badly participants felt were collected only from those who had missed a day by week 6, and that group excludes the most regular practicers. Everyone took the same free course from one service, so the cost of a broken streak may weigh differently where learners have paid. A follow-up will test whether weekly rather than daily targets serve participants with irregular work or caregiving schedules better than either streak design.

Editorial Note

This report underwent CfAHC's Rapid Review Protocol, including AI-assisted literature cross-checking and a two-reviewer human sign-off.