Module five’s “fuse date” on the predictive tail is now a footnote on every UA slide we send upstairs. I still think week two’s homework assumed more finance vocabulary than our PM had.
Ploy, product manager · streaming app
Flagship program
Cohort LTV Mastery
Six weeks for operators who already have events and still cannot defend lifetime value when finance asks a second question. Taught in English from Ratchathewi, with remote seats for teams elsewhere in Southeast Asia.
Modules
What the six weeks actually cover
-
Defining LTV without blended averages
User, clock-start, revenue basis, and lifetime horizon written into a two-page glossary. We kill the single-tile habit on day one.
-
Cohort construction for subscription and hybrid IAP
Install week versus first purchase versus plan change. Platform splits. Thai payment rails that do not behave like a US card vault.
-
Survival, censoring, and incomplete weeks
Why treating a user who installed on Thursday as a “zero LTV” observation poisons the mean. Simple survival views without pretending you run a biostats lab.
-
Margin after refunds, VAT, and store fees
Gross-to-net walk-throughs using App Store, Play, and local processors. Refund windows that actually apply to apps selling from Thailand.
-
Predictive tails — and when to stop trusting them
A cautious fit, an explicit fuse date, and a list of failure modes. No production bidder, no feature store.
-
Board-ready narrative and spending rules
A decision memo, a payback kill-rule, and a one-page curve story. Pretty dashboards remain optional.
Outcomes
You leave with artefacts, not slogans.
- A signed glossary (product + finance, even if they still disagree on taste)
- Cohort tables for at least two cuts that matter to your app
- Net contribution curves with assumptions listed in plain English
- A spending rule your UA lead can follow without a new meeting
- A memo draft marked by the instructor
Informational fee
Ledger Bench seat: ฿41,800. Vault Circle (same syllabus plus 1:1 critique): ฿96,000. Glimpse recordings are a separate, lighter product. Nothing is charged on this site — see studio fees and refunds.
Instructor
Arisa Thongchai
Former head of insights at a Bangkok super-app vertical. She has marked every Ledger Bench cohort since 2021 and still refuses to let blended LTV into the final memo.
FAQ
Before you enquire
Do I need SQL?
You need to be able to export a user-level or day-level slice, or to sit next to someone who can. We work in worksheets and critique, not in a warehouse IDE.
Is there a real limitation I should know?
Yes. This studio does not cover deep SQL engineering or building a production machine-learning pipeline. Analysts who need Spark-scale feature stores should pair this with an internal data-engineering track. We will not pretend six weeks replaces that.
Can two people share a seat?
Team seats (two people, one export) are quoted after you write us. Critique time does not automatically double.
What if our taxonomy is a mess?
Week one will show it. If the holes are too large, we pause modelling and help you write an event patch list. That can mean you finish the memo later than week six — we would rather that than a confident fiction.
From the room
How seats described the work
★★★★☆
Arisa marked our glossary until VAT and store fees were in the same sentence. Remote audio in the Ratchathewi room was occasionally thin; the worksheets were not.
Client in e-commerce checkout · Ho Chi Minh City