Learning outcomes
- A signed definition of your unit of use
- A sampling plan that keeps rare checkout and payment paths
- Alert copy your on-call will not silence in week two
- A one-page ledger suitable for a CFO conversation
Flagship course
The Usage Ledger
Eleven weeks on API Usage Analytics for Apps. You leave with a written unit of use, a sampling story that keeps rare paths, and a one-page ledger you can defend in a finance meeting.
Modules
What we cover, in order
-
What usage actually means
Calls versus unique users versus billable operations. Writing the sentence finance will sign, and listing lookalike events that must not count.
-
Instrumentation without drowning the app
Where to emit, what labels are allowed, and the calorie count of a client SDK. We refuse “log everything and filter later.”
-
Sampling, aggregation, and cardinality traps
Head versus tail sampling, histogram buckets, and the moment a well-meant user_id label explodes the bill.
-
Quotas, budgets, and alert design
Calendar-aware bands, burst behaviour, and copy that names the customer. Muted channels are treated as design failures.
-
Reading dashboards with product intent
Grading panels on whether a PM can explain cost movement. Deleting vanity charts is graded work, not optional hygiene.
-
Thailand and SEA in the graph
Bangkok peak hour, festival weekends, and regional routing. Why a US lunch-break baseline lies after 18:00 here.
-
Capstone: instrument a mobile + backend pair
A provided sample app. You ship a usage ledger, not a slide deck. The room reviews it like a finance partner would.
Informational tuition
Public Telemetry Studio seat: ฿24,800. Cohort Plus Lab (studio + private office hours): ฿33,500. There is no checkout here. See all tuition and the refund policy.
Instructor
Arthit Khemkhaeng led telemetry for a Bangkok ride-hailing platform before joining the studio. He still keeps a small on-call rotation so the lab traces stay dishonest in the useful way.
FAQ
Questions we actually get
Do I need to be an SRE already?
No. Product engineers and analytics leads do well if they can read a JSON payload and argue about product behaviour. Pure executives without a technical partner tend to struggle in weeks two and three.
Will you instrument our production mesh for us?
Not in the public studio. Production data stays yours. Cohort Plus Lab office hours can review redacted traces. In-house Desk is the contract if you want us inside the company.
What is a real limitation of this course?
Eleven weeks will not replace an observability platform migration, nor will it teach incident command. Students who arrive hoping we will pick their vendor for them leave disappointed — we compare trade-offs, we do not close a purchase. The capstone app is also smaller than a super-app; if your cardinality problem lives in a 400-service mesh, you will still need internal follow-up.
Is the recap enough if I miss studio night?
The recap covers the argument, not the interruptions. Missing more than two nights usually shows in the capstone. We would rather you defer a cohort than ghost half the table.
From this syllabus
Reviews tied to The Usage Ledger
The capstone review was more useful than the lectures. Having Arthit ask “would finance sign this unit?” in front of fifteen people is uncomfortable and exactly why I enrolled.
★★★★☆Sampling week changed our Android SDK. The SEA latency module felt thin if you already run POPs in Singapore — still worth hearing once.