Studio
We opened a room on Chakrphong Road because dashboards were outrunning judgement.
Logicfeaturegrid exists for teams who can already plot a retention curve and still cannot agree whether last quarter’s campaign worked. We teach the experimental layer: holdouts, leakage, and language that survives a sceptical review.
Mission
Our brief is narrow. We want fewer fake wins in Retention Experiment Analytics. That means teaching people to pre-register a stop rule, to respect a holdout even when the growth lead is impatient, and to describe uncertainty without hiding behind a single percentage.
We work from Thailand because this is where several of our instructors already ran marketplace and wallet experiments — and because Bangkok’s calendar (Songkran, Chinese New Year spillover, monsoon logistics) exposes sloppy windows faster than a textbook example ever will.
Pedagogy
Bring a live question. Leave with a signed decision rule.
Classes are not slide recitals. Each seat arrives with a product question and, ideally, a warehouse extract. We sketch the unit of randomisation on paper before anyone opens a notebook. Homework is a memo, not a quiz.
Live tables are capped at eleven so the same instructor who taught the evening still marks the work within seventy-two hours. Recorded materials exist as Field Notes, not as a substitute for the room.
We assume you can query events. We do not teach SQL from zero, Python from zero, or how to buy ads. If that is your gap, we will say so on the first call rather than stretch the syllabus.
Community
A quiet alumni thread, not a noisy forum.
Graduates of the live benches can join a private thread used for holdout edge cases, holiday calendar notes, and occasional visiting critiques. It is intentionally small. We do not sell access separately, and we do not promise introductions to investors.
Once or twice a year we host an open evening in Pranakhon: one case, one argument, drinks after. Seats are announced on the journal, not through urgency timers.
Student journey
From first email to a memo your VP can reject honestly.
- Write to us with the product surface and the decision you actually need to make. We reply within two working days.
- If the flagship bench fits, you receive the syllabus, the fee (informational), and the next table date. If it does not, we say so.
- Before week one you send a one-page brief: metric, population, known leakage. We comment once.
- Six teaching sessions, homework memos, and a final holdout map reviewed in person or on a scheduled call.
- You leave with artefacts, not a badge. Alumni may return for Holdout Atelier hours if a later experiment goes sideways.