
Track overview
About this track
Decision practice for the discipline that sits between shipping and learning: running experiments and deciding from them. An experiment result is not a decision, the decision is a rule that weighs the effect, the certainty it is real, and the cost it imposes on everything you were not measuring. Learn the rule: ship only what is real (you can trust the effect), meaningful (it clears a bar worth the complexity) and guardrail-safe (it does not quietly break something you protect), and, when it is not shippable, whether the honest move is to hold and fix or to kill. When running an experiment is nearly free, the scarce skill is deciding from one honestly.
How each module works
Every module in this course runs the same three steps, from the first lesson to the debrief. The loop is what makes Decyde a place to practise, not just to read.
Step 1
Read
The lessons. Read the situation, the evidence and the shape of the trade-off before you commit.
Step 2
Size
The simulation. Put numbers on both sides of the trade-off, then make the call against the engine.
Step 3
Defend
The debrief and report. See what your call bought and why, so you can defend it or change it.
The same loop in every module, and a new decision each time. Judgment compounds.
Modules · In order
4 modules in this track
Each module is one decision: short readings, the decision itself, then a debrief and a replay in a shifted world.
01/04
The Ship Call
You are Maya, Senior PM at NorthPeak, a B2B SaaS platform. An onboarding experiment reports a 15% activation lift and the team wants a Thursday rollout. The primary metric is only part of the evidence. Compare the confidence behind the result with effects elsewhere in the business, then choose whether to ship the change, hold it for a fix and re-test, or revert it. The call belongs to you.
13 lessons · 2 mini-sims · ~50 min
Begin module02/04
The Underpowered Test
You are Maya, Senior PM at NorthPeak. Last quarter you held the onboarding redesign because a retention guardrail broke under the headline. Engineering fixed it, the clean re-test closed on day 42, and it came back flat: activation 40.00% against 42.00%, a +5.00% relative lift at confidence 0.62, under the 0.80 bar. The room has already written the post-mortem. Nobody has asked what this test could have detected. On the bench: a measuring scale with the Floor across it as a band of frosted glass, the brass bar still in its cradle, a crank, a tray of five lenses and four doors. Drag the bar on, turn the crank, seat a lens - none of it commits anything, and all of it is optional. What you find is that the smallest effect this test could ever resolve is 14.00%, against a bar of 5.00%: the number everyone is quoting sits under the glass. Wind the crank to the end of the quarter and it stops at 6.60%, still above the bar - the clock does not arrive. Seat the 24-hour lens and the fog lifts to 4.05% on data already collected, with no extra weeks at all. Then roll the change through one of four doors: call it, extend, re-instrument, or ship anyway. Discover why re-instrumenting beats the whole quarter (+$61.90k against -$90.00k), why the patient reflex costs nearly two and a half times what the impatient one does, and why the identical flat number is worthless against a 5% bar and decisive against a 20% one.
13 lessons · 2 mini-sims · ~55 min
Begin module03/04
The Trustworthy Readout
You are Maya, Senior PM at NorthPeak. The re-run on the instrument you chose last quarter has closed, and it is the best result of the year: 24-hour setup completion up 9.2% relative on 20,000 users, a tight interval, every guardrail inside tolerance. You run this course's two checks and both return green. The effect clears the ship bar by 1.84 times, and the test could resolve an effect of 2.59% against a 5.00% bar. The readout is still not data. On the bench: two glass columns holding the assigned users side by side, and they look identical. Engage the balance and the gauge at the fulcrum reads 5.94 sigma, because the split was 10,420 against 9,580: a 2.1-point tilt that reads as rounding and that a fair coin does not do. Draw the rake across a pre-assignment attribute and one band does not match. Grip it, lift it out of both columns, and everything moves at once: the beam settles, the gauge falls to zero, the columns mirror, and the headline drops from 9.2% to 6.4% in your hand. The number gets smaller and better, and it still clears the bar. Then roll the readout through one of four doors: accept, discount, salvage, or void and re-run. Discover why accepting is the worst door here and the one a green dashboard invites, why a hedge that shades the number cannot change its sign, and why the identical 5.94 sigma reading demands salvage in one world and a re-run in another.
13 lessons · 2 mini-sims · ~55 min
Begin module04/04
The One Question
You are Maya, Senior PM at NorthPeak. The re-run closed clean: the arms are level, the instrument was adequate, and the primary metric, 7-day active use, came back flat. Then the room scrolls. Below the primary sit four secondary metrics and nine segments, and two rows are green: a 140-per-arm segment up 18.00%, and, on the full sample, teammate invites up 5.60% at p = 2.2243%. The segment is easy to reject. The invites row passes every check this course has taught you, and it is still not a result. On the bench: the deck in a brass rail, a census plate, a correction bar, a floor strip, and a solitary slot that reads one card as one question. Then finish the line that goes in the quarterly deck, "In this test, the change ...": claim it, confirm it, bank it, or bin it. Discover why both of the room's instincts lose, why the same p-value is 12.15 times more likely to be an accident as one of 14 questions than as one of one, why the honest correction costs resolution, and why the identical number is the best sentence in one world and the worst in another. The fourth and closing module of the course.
13 lessons · 2 mini-sims · ~60 min
Begin moduleAuthored on Decyde Practice loop curated for decision discipline
Vocabulary
What you'll engage
The decision skills you'll practice are paired with the underlying patterns you'll learn to recognize. Both vocabularies compound across the Product Decisions domain.
Decision skills you'll practice
Scope cuttingRoadmap trade-offsBuild-vs-buyCustomer-evidence weighingSequencing judgmentMVP defining
Patterns you'll see
Practice evidence recorded on Decyde
Every decision practiced builds your Decision Profile
Replays supported
Practice the same decision differently · iteration builds judgment
English
Additional languages on v1.3+ roadmap
What's next · One recommendation
The other Product Decisions courses each practise a different kind of decision.
Begin practice →FAQ
Questions about this track
Common questions about Experimentation & Evidence Decisions and how this track works.
Related · Other tracks





