★★★★☆
The holdout module caught a bug in how we counted reinstalls. I still wish session three had more SQL examples for our warehouse dialect.
Flagship course
Signal Stack Studio
A five-module cohort that turns personalization ideas into measurable app experiments — without pretending the data warehouse is already perfect.
What you will leave with
Concrete artifacts, not generic frameworks.
A
Versioned event contract
A living sheet that maps client events to analysis fields, with owners and deprecation rules.
B
Holdout checklist
Exposure rules that survive mid-cohort flag flips and silent SDK retries.
C
Readout template
A narrative structure for product leads that separates lift from composition effects.
D
Limitation memo
Practice naming what the study cannot claim — including sample bias in regional apps.
Modules
01
Surface map & intent
Inventory personalized surfaces, define success without vanity engagement, and pick one release to instrument deeply.
02
Taxonomy that survives renames
Design event names, properties, and identity joins that mobile and web can keep in sync.
03
Exposure & holdouts
Allocate users, handle late arrivals, and document when a flag change voids a comparison.
04
Cohort reading under noise
Segment carefully, watch for device mix shifts, and separate novelty from lasting lift.
05
Stakeholder readout
Write the brief, run a critique circle, and rehearse answering “what would make you ship or kill this?”
Instructor
Mira Chaiwat
Former lead analyst for a Southeast Asia marketplace app. Mira focuses on personalization measurement for ranking and messaging surfaces, and still reviews alumni experiment briefs after each cohort.
Informational pricing
Signal Stack Studio is listed at THB 24,900 per seat. Team bundles appear on the plans page. This site does not process payments — contact us to reserve seats.
“We used the readout template the week after class. Our eng manager stopped asking for a single conversion number.”
FAQ
Do I need a production ML ranking system already?
No. The course assumes you personalize something — feeds, messaging, onboarding — and need trustworthy measurement. Deep model training is out of scope.
What is a real limitation of this course?
We do not certify your company’s compliance posture, and we cannot guarantee lift numbers. If your SDK cannot emit stable IDs, Module 02 will surface that gap rather than paper over it.
How much time outside live sessions?
Plan roughly four focused hours between sessions for audits and write-ups. Teams sharing one seat should designate a scribe.
Is payment available on this site?
No. Pricing here is informational. Reach out via the contact form to discuss invoices and cohort dates.