“I came for dashboards and spent the second week arguing about whether first-open or first-purchase starts the clock. Ann would not move on until we wrote it down. That argument is now a one-page policy our CPO actually quotes.”
Reviews
Not every evening was comfortable. The tables still shipped.
These notes come from people who sat Window Clinic, Retention Windows Atelier, or Cohort Residency. We keep a mild reservation in the room on purpose.
★★★★☆
The paywall return-curve module was the piece I used immediately. I still think the Thailand seasonality evening could have used one more local dataset — we leaned hard on Songkran and year-end, and my category peaks in July.
Window Clinic was exactly one Saturday. I walked in with a broken Day-14 and left with a denominator I can explain to finance.
★★★★★
Residency desk hours were blunt. My stakeholder readout rehearsal failed twice. The third version survived the board pack.
English instruction was a relief for our mixed team. Just know: if your pivot-table muscle is weak, you will feel behind before module three of Day-7 Survival Maps thinking, even though that short course is gentler than the Atelier.
I wanted more visualisation craft. Proxy Cloudhub will not give you that. They will make you name the window.
Case notes
Two rooms, two broken windows
Songkran looked like a retention collapse
A marketplace team treated a Day-7 drop during Songkran as a product failure and paused a checkout experiment. In Window Clinic we rebuilt the cohort with acquisition stamps shifted to “first successful order” instead of first-open, and excluded the holiday week from the comparison set rather than from the product narrative.
The “collapse” was mostly people leaving Bangkok. The experiment resumed. The lasting change was a written rule: holiday windows are annotated, not silently deleted.
Trial start was not first-open
An audio app mixed trial-start with first-open inside the same Day-14 return curve. Paid conversion looked healthy; returning listeners after trial looked mysteriously weak. During Retention Windows Atelier the student split the clocks: one window for trial, one for first content play after subscribe.
The mild cost: two extra columns and a longer readout. The gain: paywall copy stopped being blamed for a clock-start error.