Abstract frameworks are useful. Watching one work in practice is more useful.
This is an eight-week account of one person — Marcus, a 34-year-old product manager — using the DETACH Method and AI check-ins to reduce a persistent late-night phone-scrolling habit. Marcus agreed to share his check-in notes and reflections on the condition that his last name not be used.
The account is not cherry-picked success. Week 3 was a failure. Week 5 nearly produced a full relapse. The overall arc is positive, but the honest middle is what makes it worth reading.
The Starting Point: What the Habit Actually Looked Like
Marcus had been aware of the habit for about two years. His description at the start of Week 1:
“I get into bed between 10:30 and 11. I tell myself I’ll check a few things and then sleep. What actually happens is I’m still scrolling at midnight or later, sometimes 1am. I wake up tired. I know I’m doing it, I don’t want to do it, and I can’t seem to stop.”
His initial theory about the trigger: boredom.
After the cue detection conversation in Week 1, the actual picture was more specific. The AI asked him about the last seven instances in detail. The pattern that emerged: the behavior was most likely to extend past midnight on days when work had been stressful and unresolved — specifically, days when he’d left a difficult conversation or an unfinished project without closure. The phone wasn’t filling boredom. It was providing escape from an unresolved, uncomfortable mental loop.
The HALT classification: primarily Tired (it was always late in the day) and sometimes Angry (residual frustration from work conflicts). The boredom hypothesis was wrong.
Week 1: Cue Detection and First Friction
Marcus ran the cue detection conversation on a Tuesday. By Friday, he had implemented two environmental changes:
- Phone charged in the hallway starting at 10pm, using an extension cord that made it too short to reach the bed
- A notepad on the nightstand for capturing the work thoughts he identified as the actual driver
The notepad was his addition — unprompted by the framework. He reasoned that if unresolved work loops were the trigger, having a place to offload them before bed might address the need. This proved to be one of the most effective interventions in the eight weeks.
Week 1 check-in result: 5 out of 7 evenings involved late-night scrolling (down from an estimated 7/7 prior). The two improvement nights were both nights he’d used the notepad.
The AI’s response to the Week 1 check-in included this observation: “The notepad correlation is worth paying attention to. Before we assume the phone distance is working, let’s see if the two things that changed were the phone location or the notepad. Can you test them separately next week?”
Week 2: Testing the Variables
Following the AI’s suggestion, Marcus tried the notepad without moving the phone for three days. On all three days, the scrolling occurred, though the duration was slightly shorter.
He then moved the phone without the notepad. The phone distance alone produced no reliable improvement.
On the two days he used both: no extended scrolling.
The conclusion: the underlying driver was the unresolved work loop, not proximity to the phone. The notepad addressed the need; the phone distance prevented the default behavior from filling the gap.
Week 2 check-in result: 3 out of 7 evenings involved extended scrolling. The notepad was the critical variable.
Week 3: The Failure Week
Week 3 was Marcus’s worst work week in months. A product launch was delayed, a key engineer quit, and he had a difficult conversation with his manager that he felt he’d handled poorly.
By Thursday, the habit had fully reasserted. By the weekend, he’d stopped using the notepad. By Sunday’s check-in, he was down to one successful evening out of seven.
His self-assessment in the check-in: “I basically gave up this week. The whole thing feels pointless.”
The AI’s response avoided the reassurance trap. Instead:
“Week 3 is actually very useful information. You’ve confirmed that the cue is work-stress, not just tiredness — because this was an unusually high-stress week and the behavior came all the way back. That means the notepad works when stress is average, but not when stress is very high. The question is whether there’s a version of this approach that works even in hard weeks — or whether we need to accept that hard weeks will look different and focus on recovery speed rather than prevention.”
This reframing — hard weeks are data, not verdicts — was the turning point. Marcus described it as “the first time I felt like the system was mine rather than something I was failing at.”
Week 4: Rebuilding After the Failure
Week 4 involved three changes:
- A modified notepad protocol for high-stress days — a more structured format that included not just the work loops but also a “what I’m going to let go of tonight” section
- The phone charger back in the hallway, with an additional blocker app on the social media platforms most likely to extend the sessions
- A HALT check at 9:30pm — a 2-minute inventory of his state before getting into bed
The 9:30pm HALT check was new. The AI suggested it based on the Week 3 data: if high-stress states are the primary driver, catching the state before the trigger fires gives more response time than trying to intervene in the bed at 11pm.
Week 4 check-in result: 4 out of 7 evenings without extended scrolling. The worst night was Wednesday — a day he’d skipped the 9:30pm check.
Week 5: The Near-Relapse
A different kind of challenge arrived in Week 5. Marcus had a run of easier work days, and his motivation for the process dropped. He stopped using the notepad for four days — not because he was stressed, but because things were going fine and the discipline felt unnecessary.
By Friday, the extended scrolling was back — not from stress but from simple habit reassertion. When the friction and replacement routine dropped, the original behavior refilled the space.
This is a documented pattern in habit change research. The middle weeks often include a “motivation dip” when early progress has been made and the new behavior isn’t yet fully automatic. The system needs to sustain the change even when motivation is absent.
Week 5 check-in result: 3 out of 7 evenings without extended scrolling.
The AI observation: “This week is different from Week 3. Then the habit came back because stress was high. This week it came back because the structure wasn’t in place. That’s actually easier to fix — the structure is under your control. The question is whether you’re willing to run the routine even on easy days, which is when it feels most unnecessary.”
Marcus’s reflection: “I realized I was treating the routine as something I needed when things were bad. The reframe was that it’s a system, not a crutch — it works because you use it, not just when you feel like you need it.”
Weeks 6 Through 8: Consolidation
The final three weeks showed a steadier trend. The combined protocol — 9:30pm HALT check, structured notepad, phone in hallway, blocker app — became more automatic.
Week 6: 5/7 evenings without extended scrolling. Week 7: 5/7 evenings without extended scrolling, with the two slips both on Friday nights (a social pattern the AI flagged as worth noting). Week 8: 6/7 evenings without extended scrolling.
The Friday pattern led to a new question: what was different about Fridays? The AI check-in surfaced it: Marcus had been using Friday evenings as deliberate decompression nights — “I’ve earned this” — which reintroduced the extended scrolling as a reward rather than a stress response. This was a different cue structure that the original protocol hadn’t addressed.
The adaptation: a deliberate “wind-down period” on Friday evenings that included the phone but was time-limited and intentional — rather than the unintentional, extended late-night version. This preserved a sense of relaxed evenings while maintaining the structure.
What the Eight Weeks Showed
What worked:
- Cue detection was the highest-leverage investment. Thirty minutes of honest conversation revealed the actual driver (work loops) vs. the assumed driver (boredom). This changed the entire approach.
- The notepad replacement addressed the need. Environmental friction alone didn’t work.
- The 9:30pm HALT check as a prophylactic — catching high-vulnerability states before the cue window — was more effective than trying to intervene at the moment of the cue.
- AI check-ins surfaced the Friday pattern and the Week 3 stress-vs.-routine distinction. These patterns would not have been visible without cross-session reflection.
What didn’t work:
- Phone distance alone without addressing the underlying need.
- The assumption that “good weeks” don’t require the structure.
- The initial binary framing — either the habit is “broken” or it isn’t.
What Beyond Time added: Marcus used Beyond Time’s planning workspace from Week 4 onward to log his check-in notes and track the habit alongside his weekly planning. The persistent context meant the AI referenced the specific patterns from previous weeks without Marcus having to re-explain them. He described this as “the difference between a check-in and a continuity.”
The Honest Numbers
| Week | Evenings without extended scrolling | Key event |
|---|---|---|
| 1 | 2/7 | Cue detection, first friction |
| 2 | 4/7 | Notepad identified as key variable |
| 3 | 1/7 | High-stress week, full relapse |
| 4 | 4/7 | HALT check added, protocol rebuilt |
| 5 | 3/7 | Motivation dip, structure abandoned |
| 6 | 5/7 | Protocol stabilizes |
| 7 | 5/7 | Friday pattern identified |
| 8 | 6/7 | Friday protocol adapted |
The overall trend is real. The middle weeks are genuinely hard. The slips are informative. This is what habit change actually looks like — not a smooth downward line, but a jagged arc with setbacks that become data.
For the full framework Marcus used, the DETACH Method guide covers each stage in depth. The Beyond Time walkthrough shows the specific features Marcus used for tracking and check-ins.
Your action today: If your habit-breaking attempts have involved a week like Marcus’s Week 3 or Week 5, ask yourself: did you treat that week as failure or as data? The answer tells you a lot about whether your current approach can sustain through difficult periods. If the answer is failure, the DETACH framework gives you a different lens.
Frequently Asked Questions
-
How was success measured in this case study?
Three metrics were tracked: frequency of the original behavior (number of evenings with more than 30 minutes of late-night scrolling), sleep onset time, and subjective wellbeing the following morning. All three were self-reported through weekly AI check-ins. This is observational self-report data — not controlled experimental data. The value is in the pattern, not statistical proof.
-
What made the AI check-ins different from journaling?
The key difference was the AI's ability to reflect patterns back across sessions. A journal entry captures a moment; the AI check-in referenced previous weeks and asked about changes and trends. This longitudinal reflection surfaced patterns (the Thursday vulnerability, the correlation with late work days) that wouldn't have been visible in isolated weekly entries. The non-judgmental tone also enabled more honest reporting than most people manage in personal journals.
-
Did the habit fully disappear?
No, and that's an important part of the honest account. By week 8, the behavior had reduced from a near-nightly occurrence to roughly one to two times per week, and the duration was significantly shorter. The reduction is meaningful and sustainable. Complete elimination of the behavior in all contexts isn't a realistic goal for most habit change — and treating partial success as failure is one of the patterns the framework is specifically designed to prevent.