B049 · Inference Under Uncertainty

Planning fallacy

Underestimating time and cost by relying on the inside view rather than base-rate forecasting.

Severity

High

Frequency

High

Decision stages

Decide

Business impact: Produces chronic project overruns and missed delivery commitments.

What it is

The planning fallacy is the systematic tendency to underestimate how long a task will take and how much it will cost, while overestimating the quality of the outcome. It occurs because planners focus on the specific sequence of steps required for this project - the inside view - rather than asking how long comparable projects have actually taken. The inside view generates optimistic scenarios; the outside view, grounded in base rates, consistently corrects towards longer timelines and higher costs.

Where it shows up

In custom software build decisions, the planning fallacy is almost always present. Teams estimate from their own plan rather than from the distribution of outcomes for comparable builds. Integration complexity, testing cycles, stakeholder change requests, and operational onboarding are consistently underweighted. A six-month estimate for a bespoke build frequently reflects a best-case scenario rather than the median or realistic case - which is the number that should drive the investment decision.

What Rubicon Probity does

When Rubicon Probity reviews a decision record at the Decide stage where build estimates are presented without reference-class forecasting, it raises a CAUTION flag and requests an outside view alongside the inside view. The record must document a comparable project class, the median outcome from that class, and an explicit risk-adjusted estimate before the business case is approved.

Detection questions

  • Has this estimate been produced using only the project plan, or has it been tested against the distribution of outcomes from comparable builds?
  • What is the median delivery timeline for the reference class of projects most similar to this one - and by how much does your estimate differ from that median?
  • Have you modelled the P50, P80, and P90 scenarios, or only the plan that assumes everything goes to schedule?
estimationproject-managementbudgetingforecastingbuild-vs-buy