What the room said after the tables were honest

• Retention diagnostic

  The diagnostic caught that our session_start fired on every notification tap. We had been celebrating Day-7 numbers that were mostly people dismissing an alarm. Aina was blunt about it, which I needed. The written notes were slower than I hoped — we waited nine days for the first cohort table — but once they arrived the argument inside the team stopped.

• Cohort construction workshop

  They made us cut cohorts by first completed recitation, not by install date. That single change showed the drop after Ramadan clearly. We had been averaging it away.

• Habit-loop review

  Priya’s habit-loop review found three reward events we never logged — the save-to-journal tick, the shared verse, the teacher stamp. Useful. I still think they underweighted our school-term calendar; Malaysian classroom apps do not behave like imported consumer products, and we had to push that point in the readout.

• Standing retention advisory

  The monthly session is short on purpose. Wei Jun noticed we were counting prescribed rest days as streak breaks. That would have been an ugly reminder to send to people following a clinic plan.

A longer visit — Penang stretch product

When Day 7 was mostly dismissed alarms

Mei Lin’s team had a morning stretch circuit and a 6:30 reminder. Their install-week Day 7 looked like a product that had found its people. In the diagnostic we recut week zero as the first completed ten-minute circuit. We also marked session_start as a cue event after watching it fire on a notification tap that never opened the mat view.

The new Day 7 among people who had actually stretched was lower. The following two weeks among those same people were steadier than the old table suggested. The team spent the working session renaming events and moving the reminder copy so it no longer claimed a session had begun. They did not leave with a new acquisition plan. They left with a glossary they could defend in their own meeting.

The nine days to the first table was slower than they wanted. We still will not rush a canvas to match a board date.

A longer visit — Johor drill product

Teacher stamps that never entered the log

Shalini’s after-school drill product had a generous first-lesson completion rate and a quiet second week. The habit-loop review found three rewards the classroom already used — a journal tick, a shared verse, a teacher stamp — none of which wrote an event. Return tables could not see the thing teachers cared about.

The reservation in her note is fair. We first cut festival weeks too coarsely for a school calendar. She brought term dates to the readout; Wei Jun recut the month. The glossary change still mattered. So did the calendar she insisted we put on the canvas.

If your argument is stuck on a number you do not trust, write with the ritual and the week people vanish.