We keep the glowing lines, and we keep the ones that sting. All of these refer to programmes currently on this site.
★★★★☆
Activation Window Design forced us to admit our “aha moment” was a screen nobody reached until day four. We moved the prompt. Install-to-first-value stopped looking like a personality flaw in our users.
Came for Mixpanel tricks. Left with a list of seventeen events to delete. Slightly annoyed about that, in a useful way.
The weekly readout template from Retention Signal Architecture is now the only analytics slide our board sees. I still think the homework volume in weeks 4–6 is heavier than the syllabus admits. Budget the evenings.
House Briefing, two days in our Shoreditch office. Amira refused to look at the growth dashboard until the event names were readable aloud. Our CTO found that theatrical. By lunch on day two he was renaming things himself.
★★★★★
Store Revenue Telemetry did not make App Store Connect pleasant. It did stop us double-counting family-plan upgrades. That paid for the Reading Desk twice.
Case note 01
A language app that retired its streak as a north star
A 22-person team in Glasgow joined Studio 12 with a streak metric on every all-hands slide. Streaks were rising while paid retention at day 35 was not. During the cohort-honesty module they rebuilt the curve for learners who completed one lesson in week one versus those who only opened the mascot screen.
The mascot-openers had been inflating “engaged users.” They moved the activation definition to “first lesson with a spoken reply.” Six weeks after the studio they reported that trial messaging now points at that lesson, not at the streak. They did not claim a miracle; they claimed a less embarrassing slide.
Case note 02
A neobank that stopped arguing about “active”
Three squads, three definitions of an active customer. The House Briefing spent the first morning writing the disagreements on paper and refusing to open a laptop. By the afternoon they had a single verb — “completed a customer-initiated payment” — and a 7-day window that matched salary cycles in the United Kingdom.
The follow-up office hour, six weeks later, was mostly about what to tell Marketing, who had liked the old, larger number. We sat on that call. The number got smaller. The argument got shorter.
