Bangkok studio · iOS & Android
Read the test. Leave the theatre.
Systemtracebase is a small school for Mobile A/B Test Analytics. We train product, growth, and analytics leads to design experiments that survive a real release train — store listings, permission prompts, paywalls, and the quiet metrics that tell you a “win” is just seasonality.
From the floor
“An Android holdout is not a mood. If you cannot name the guardrail before you name the variant, you are not running a test — you are decorating a release.”Nalinee Prasert · Lead instructor
Studio catalogue
Three desks, one flagship
Experiment Design for Mobile Releases
Hypothesis writing, unit of randomisation, and sample plans that respect app-store review cycles.
Funnel Wiring for App Analytics
Event taxonomies, identity stitching, and the places mobile SDKs quietly drop users.
Power, Peeking, and Guardrails
How to stop a test early without lying to yourself, and which metrics may never be primary.
How the room works
We teach on live release notes, not slide decks alone.
Each Variant Atelier week opens with a real experiment brief from a past cohort or a public app. You write the hypothesis, pick the randomisation unit, and defend the sample size before anyone opens Figma. Critique is spoken, timed, and written back to you the same evening.
Bangkok office hours sit on Visuthkasat Road. Remote seats join the same call; we do not run a parallel “online-only” curriculum that skips the hard reviews.
What you leave with
Skills that show up in the next sprint
A sample-size habit
You will stop launching 4% lifts on 1,200 weekly actives and calling it science. We practice MDE selection against actual Thai and SEA traffic profiles.
Store-aware variants
Listing screenshots, in-app paywalls, and permission copy are different surfaces. The course treats App Store review delay as a design constraint, not an afterthought.
Readable readout memos
Every graduate writes a one-page experiment memo: decision, uncertainty, and what will not change in the next release. Stakeholders get the memo, not a dashboard dump.
Guardrail literacy
Crash-free sessions, unsubscribe rate, and support tickets sit beside conversion. We mark which ones can veto a “winning” variant.
Voices from the desks
What people actually say after week eleven
The randomisation-unit lesson in Experiment Design for Mobile Releases stopped us splitting users by device ID on a shared family plan. Painful to unwind, but we caught it before the Deepavali campaign.
I wanted more iOS StoreKit lab time than the syllabus allowed. The Funnel Wiring evenings still paid for themselves: we found a double-fire on checkout_start that had been inflating every paywall test since March.
★★★★☆
Clear on peeking rules. The Variant Atelier critique is blunt. If you need cheerleading, this room will feel cold.
Journal
Notes from the reading circle
Sample size planning when your weekly actives wobble
Thai festival weeks, payday spikes, and why a flat MDE calculator will embarrass you in October.
How three Thailand teams ship a weekly mobile test without a data science bench
A look at release-train habits from Chiang Mai, Phuket, and our own Bangkhunprom alumni.
On Visuthkasat Road
Sit in on a critique before you enrol.
Write to us with the experiment you are currently arguing about. If the timing works, we will invite you to a Wednesday readout. Pricing is listed; there is no checkout on this site.
Street: 371-373 Visuthkasat Road Bangkhunprom; City: Pranakhon; State/province/area: Bangkok; Zip code: 10200; Country: Thailand
Phone: 662 2825668