Decision-making under uncertainty
On this page 9 topics
Overview
Topics
- Assumptions
- experiments
- option value
- decision records
Learn first
3 h of practice (estimate)
Learning outcomes
- Make assumptions explicit and rank them by impact and uncertainty.
- Design cheap experiments that reduce the most important uncertainty.
- Value keeping options open and decide when to commit.
- Record decisions with their assumptions so they can be revisited when assumptions fail.
Lessons
Lab and exercises
Lab
SRA-013-LAB1 · Marketplace bets
Outcomes: 1–4 · Time: ~2 h
- Build an assumption map for the Commerce 2030 marketplace architecture with at least fifteen assumptions, ranked by impact and uncertainty.
- Design two experiments for the top assumptions with thresholds, cost caps, and decisions each outcome triggers.
- Identify which decisions to commit now and which to keep open, and why.
- Write decision records that include assumptions, signals, and review triggers.
- The assessor invalidates one assumption mid-lab; respond by finding the affected decisions and proposing options.
Acceptance: response follows from the recorded assumptions.
Exercises (case-based; six exercises)
| ID | Outcome | Task |
|---|---|---|
| SRA-013-EX01 | 1 | List ten assumptions behind the plan "launch the marketplace in three countries in nine months" and rank them. |
| SRA-013-EX02 | 2 | Design an experiment for "sellers will integrate through our API" with a threshold and cost cap. |
| SRA-013-EX03 | 2 | Critique an experiment that cannot change the decision and rewrite it. |
| SRA-013-EX04 | 3 | Decide whether to commit to one payment provider now or keep two open; justify with option value and a commit date. |
| SRA-013-EX05 | 4 | Write a decision record for the seller data store including assumptions and review triggers. |
| SRA-013-EX06 | 4 | A recorded assumption fails (bulk file uploads dominate). Identify dependent decisions and propose options. |
Show solutions
- EX01 Include business assumptions (seller count, order mix, commission), technical ones (peak, latency), and regulatory ones (residency); rank by impact and uncertainty; start with the high-high cell.
- EX02 Hypothesis: at least 60% of three pilot sellers integrate within two weeks without support. Method: pilot with documentation only. Cost cap: two engineers for three weeks. If it fails, plan a batch-file path.
- EX03 A load test run after the contract is signed cannot change the decision; move it before the commitment and set the threshold that decides.
- EX04 Keep two open behind an interface if the interface is cheap and uncertainty is high; set the commit date at contract renewal; commit sooner if delay costs exceed the option value.
- EX05 Record context, decision, options, assumptions with signals (seller write rate, data volume), consequences, review trigger (quarterly or when a signal trips), owner.
- EX06 List decisions that used the assumption (ingestion interface, capacity plan, support model); options: adapt, add batch path, reverse; compare remaining cost, not sunk cost; update records.
Outcome 3 is also covered by the lab and the quiz.
Workplace evidence: a decision you made with explicit assumptions and later revisited.
Quiz
Outcomes: O1 assumptions ranked · O2 cheap experiments · O3 option value and commitment · O4 decision records.
A self-check for you: nothing is graded, saved or sent anywhere. Choose an answer, then check it. You can try again as often as you like.
Practical part
Six case exercises SRA-013-EX01 to EX06 plus the lab with a mid-lab invalidated assumption (decision simulation). Pass: quiz at least 8/10, and a simulation where the response follows from the recorded assumptions and decisions are revisited with reasons.
Common mistakes
Analysis paralysis; decisions recorded without why.
Assessment
Decision simulation.
Workplace evidence
A decision you made with explicit assumptions and later revisited.
Career relevance
Technical due diligence (STF-010) and solution options (ARC-033).