01Success
Self Disruption
Netflix streaming transition
Netflix, Inc. · 2007–2020
How aggressively should Netflix fund Internet delivery while operating its DVD subscription business?
A field guide to consequential business decisions
Read the decision before you read the result.
54 evidence-bound cases built from filings, government records, contemporaneous reporting, and reconstructed financials—designed to train judgment without rewarding hindsight.
Archive 01Research snapshot
54decision episodes
809source records
648financial tables
1366supported claims
27 successes & turnarounds27 failures
Each case separates what a decision-maker could know then from what became clear later.
Start with contrast
Similar pressure. Different choices.
02Failure
Self Disruption
Blockbuster decline
Blockbuster Inc. · 2000–2010
How should Blockbuster allocate capital among its store network, Total Access, and nascent digital delivery while preserving financial flexibility?
09Success
Capital Allocation
Berkshire Hathaway capital allocation
Berkshire Hathaway Inc. · 1965–2024
Should Berkshire pursue the remaining GEICO shares subject to independently set price, return, reserve, liquidity, and management gates; retain its existing approximately half ownership; or allocate incremental capital elsewhere?
12Failure
Financial Forensics
Enron
Enron Corp. · 1997–2001
Should the equity committee exit, reduce to a de minimis verification hold while suspending new exposure, hold benchmark weight, or add to Enron after the Q2 filing and CEO transition?
20Success
Deep Technology Ecosystem
Nvidia CUDA platform
NVIDIA Corporation · 2006–2024
Should NVIDIA limit CUDA to graphics-adjacent and short-horizon support, sustain it as a staged external developer platform, or commit immediately to an ungated company-wide computing-platform expansion?
26Failure
Technical Validation
Theranos
Theranos, Inc. · 2003–2018
Should the board continue patient testing and retail expansion, pause for independent validation, ring-fence a narrowly validated menu, or begin a partner-controlled strategic review?
The two-part method
Commit to a view before the outcome contaminates it.
Part A freezes the evidence at a real decision date. You make a recommendation, record confidence, identify missing evidence, and set reversal conditions. Only then does Part B reveal the action, outcome, rival explanations, counterfactuals, and transfer limits.
See how the evidence boundary works01
Part A
Evidence available at the cutoff
02
Commit
Decision, confidence, conditions
03
Part B
Outcome, causes, transferable rules
Evidence, not lore
Every consequential claim earns a trail.
Tier-one sources
Filings, annual reports, court records, regulators, and government investigations.
Material-claim coverage
1366 of 1366 material claims carry structured support.
Rule hypotheses
Candidate rules retain boundary conditions and counterexamples; none masquerade as universal laws.
Choose a live decision