Flagship course

Cohort Signal Lab

Eleven weeks of critique on live-shaped exports. You will leave able to defend a trial-to-paid number in front of someone who signs the P&L — including when that number looks worse than the dashboard implied.

Ledger Studio seat · Bangkok evenings · recordings available after each critique

Cohort charts on a laptop during a Lab session

Modules

  1. Receipt anatomy across App Store and Play

    Product types, pending, family share, and what a “purchase date” actually stamps. You annotate a redacted export before any BI tool opens.

  2. Cohort construction without vanity windows

    Choose a clock and write it down. Day-0 is not day-of-install if the trial started later.

  3. Trial-to-paid leakage

    Map the four silent exits: never entered payment, entered and failed, entered and refunded, entered and charged in a different SKU.

  4. Introductory offer distortion

    Separate offer convertors from standard convertors. Mixing them is how “attach” becomes a press release.

  5. Refund and chargeback as product telemetry

    Timestamp refunds against paywall and permission changes. Build a three-reason taxonomy.

  6. SKU matrix and price-point isolation

    A short pass of the Experiment Studio checklist so Lab alumni do not contaminate a test the week after class.

  7. Dashboard critique workshop

    Bring a screenshot (or ours). We strike metrics that cannot be rebuilt from the store file. This is not a Tableau class.

Instructor

Portrait of Kritsada Phan, lead instructor

Kritsada Phan

Lead instructor. Former monetization lead for a Bangkok-based consumer subscription catalog shipping on both stores. Kritsada designed the Lab’s metric-contract template after watching too many quarterly decks invent a conversion rate that finance could not rebuild. Guest sessions on attribution windows are co-taught with Mara Ellison.

Questions we actually get

Do you cover server-to-server notification engineering?

No — and this is a real limitation. The Lab reads the notifications you already store; we do not teach you to implement App Store Server Notifications or Play RTDN pipelines. If your export is missing fields because engineering never persisted them, we will say so, but we will not write the listener. Attribution Atelier also stops short of implementation.

Can I join without sharing my employer’s numbers?

Yes. We assign a public catalog with enough mess to be useful. You will still write a metric contract as if finance were in the room.

Is the evening Bangkok time mandatory?

Critique attendance is expected twice a week. Recordings exist for travel weeks. Signal Desk is the self-paced alternative if your timezone cannot meet the room.

What tools do you require?

A spreadsheet you already trust and the store exports you can obtain. We do not require a specific BI licence. SQL helps; it is not a gate.

From people who sat this Lab

The introductory-offer module embarrassed our board slide. We had been reporting a blended convert that made the new annual look inevitable. After Lab we publish two lines. Marketing complained for a week. Finance stopped asking follow-up emails.

— Niran V., product analytics, education subscription
★★★★☆

Kritsada is precise to the point of stubborn. I wanted a prettier template for Play ladders; the Lab insisted I write the window first. Correct, still annoying on a Thursday night.

Platform-style note · catalog ops

Short version from Chiang Mai: we finally stopped treating grace as paid.