Decision Frameworks

Make better decisions faster by knowing what kind of decision you're making.

How to use

• /decision-frameworks Apply decision framework constraints to this conversation.
• /decision-frameworks <decision> Apply the right framework to the described decision.

Constraints

Decision Typing

Classify every decision before making it:

• One-way door (irreversible): pricing model, platform migration, market positioning. Take more time.
• Two-way door (reversible): feature design, copy, UI layout. Decide fast, iterate based on data.
• MUST match analysis effort to decision type. Don't spend a week on a two-way door.
• SHOULD default to treating decisions as two-way doors unless proven otherwise
• NEVER treat a reversible decision like an irreversible one. Speed matters.

Decision Quality

• MUST define the decision clearly: what exactly are we choosing between?
• MUST identify the criteria: what matters most in this decision?
• SHOULD list options with pros, cons, and unknowns for each
• MUST set a deadline. Decisions without deadlines never get made.
• SHOULD designate one decision-maker. Consensus is slow and often produces mediocre outcomes.

Information Sufficiency

• MUST decide how much information is enough before gathering more
• For most product decisions, 70% confidence is sufficient to act
• SHOULD ask: "what additional information would change my decision?" If nothing, stop analyzing.
• MUST distinguish between information that reduces risk and information that delays action
• NEVER use "we need more data" as a way to avoid making a hard call

Decision Communication

• MUST document: what was decided, why, what alternatives were considered, and who was involved
• SHOULD share decisions with affected teams promptly
• MUST create space for disagreement before the decision, not after
• Once decided: disagree and commit. NEVER undermine a decision you were part of making.
• MUST be willing to revisit decisions when new information fundamentally changes the context

Anti-Patterns

• Analysis Paralysis: endlessly gathering data to avoid committing
• Decision by Committee: requiring consensus on everything, getting mediocrity
• The Redo: making the same decision three times because it was never properly committed to
• HiPPO Decisions: highest paid person's opinion always wins regardless of evidence
• The Non-Decision: not deciding and letting the default happen by inaction