The Planning Fallacy When Comparing Options
Compare timelines with actual completion distributions, dependency risk, and rework instead of summing ideal task durations.
Example: internal build or configured service
The internal plan lists 42 workdays but ignores queue time for security review. Comparable builds took 70 to 110 elapsed days. The service has a shorter but contract-dependent range, changing the deadline comparison.
A plan is not a forecast
Plans describe how work should proceed if tasks start on time and assumptions hold. Forecasts describe the range of completion dates after ordinary variation. An option with more detailed planning can look safer merely because its unknowns are hidden inside precise tasks.
Turn plans into ranges
- Collect durations from similar completed work.
- Separate hands-on time from elapsed time.
- Map external dependencies and rework.
- Compare both options at the same confidence level.
Novel work needs wider bounds
A weak reference class does not justify a single guess. State the range, the source of uncertainty, and the milestone that will update it.
The decision rule
What to record before acting
Use observed elapsed-time ranges and shared confidence levels when comparing delivery promises.
Apply the rule in context with test the time or volume switching point. Keep the evidence and assumptions with the decision so an outcome review can distinguish reasoning from luck.
Continue within decision psychology
Loss Aversion When Choosing Between Two Options
Detect loss aversion by rewriting both options as future bundles and separating real switching costs from the discomfort of giving something up.
Loss Aversion When Choosing Between Two OptionsAmbiguity Aversion in Decisions
Reduce ambiguity aversion by giving both the familiar and unfamiliar option the same evidence tasks, ranges, and pilot opportunity.
Ambiguity Aversion in Decisions