What Is the Planning Fallacy and How Do I Fix It?
Written By Aftertone Team
Thursday, May 14, 2026
15 min read
What Is the Planning Fallacy and How Do I Fix It?
The planning fallacy is the systematic tendency to underestimate how long tasks and projects will take, how much they will cost, and how many things will go wrong, while overestimating the benefits of completing them. Identified by Daniel Kahneman and Amos Tversky in 1979, it is one of the most consistent and universal cognitive biases documented in behavioural science. It affects individuals, teams, corporations, and governments. Experience does not reliably correct it. Awareness does not reliably correct it. The only intervention with consistent research support is changing the information source used to make the estimate.
The inside view and the outside view
Kahneman and Tversky identified the mechanism precisely. When people estimate how long something will take, they use what Kahneman calls the inside view: they construct a mental simulation of the task unfolding, imagining the steps proceeding reasonably well, and estimate based on that imagined scenario. The inside view focuses on the specific case, the optimistic trajectory, and the plan as it should work.
The outside view takes a different approach. Instead of asking "how long will this take?" it asks: "what is the distribution of outcomes for projects like this, based on historical data?" It treats the current task as one instance in a reference class and uses the class's actual completion times to inform the estimate.
The inside view is more natural and more compelling. It produces more optimistic estimates. It is also less accurate. Every planning fallacy study that compares the two finds that outside-view estimates are significantly closer to actual outcomes than inside-view estimates, across domains from student projects to construction to software development to government infrastructure.
Why experience doesn't fix it
The most consistent finding in planning fallacy research is that expertise and experience do not reliably correct the bias. Buehler, Griffin, and Ross's foundational studies found this for students. Bent Flyvbjerg's large-scale analysis of infrastructure projects found it for engineers, project managers, and procurement teams with decades of experience. Roger Buehler and colleagues' follow-up research found it persists even when people are explicitly reminded of their past planning performance and asked to account for it.
The mechanism is attribution. When a project overruns, the overrun gets attributed to specific external causes: the stakeholder who changed requirements, the dependency that wasn't delivered, the technical problem that nobody could have predicted. Each overrun is explained as an anomaly, not incorporated as evidence about the base rate. The next estimate begins fresh from an optimistic inside view, without an updated model of how long projects like this actually take.
This explains why the planning fallacy is not corrected by experience. Experience provides the overruns. Attribution prevents the overruns from updating the underlying model. The model stays optimistic. The next estimate is as overoptimistic as the last.
Reference class forecasting: the fix
The intervention with the most consistent research support is reference class forecasting (RCF), developed by Kahneman and formally applied to major infrastructure project planning in Denmark, the UK, and elsewhere. The method has three steps::
Identify the reference class of similar projects.
Establish the distribution of actual outcomes for that class.
Use that distribution to anchor the current estimate, making explicit adjustments only for features of the current project that genuinely distinguish it from the class.
The critical feature of RCF is that it forces the outside view. The estimate is anchored to historical data rather than to the imagined scenario. This does not eliminate optimism entirely, but it replaces unconstrained inside-view optimism with a historically grounded baseline that includes the actual overrun rates of past projects.
Flyvbjerg's analysis of thousands of infrastructure projects found that RCF consistently produced more accurate estimates than conventional planning, with the largest improvements in projects that differed most dramatically from their initial estimates (the ones most affected by the planning fallacy). For individual knowledge workers, RCF at the personal level means tracking planned versus actual time for recurring task types and using historical completion times as the basis for future estimates rather than imagined scenarios.
The personal correction factor
A practical shortcut: once you know your average underestimation ratio for a task type, apply it as a correction multiplier. If client proposals consistently take 60% longer than estimated, multiply every future proposal estimate by 1.6. If writing tasks consistently take twice as long as planned, double every writing estimate. The correction factor is not precise, but it accounts for the systematic portion of the bias that attribution has been preventing from updating the estimate.
Two to three weeks of planned versus actual tracking generates enough data to identify the largest systematic gaps. The correction factors derived from this data are specific to your work and your context, which makes them more accurate than any general "multiply by 1.5" heuristic. The Aeon et al. (2021) meta-analysis on time management found that planning behaviours with feedback loops, where actual outcomes are tracked and used to update future plans, produce significantly better performance outcomes than planning without feedback.
Task breakdown as a partial intervention
Decomposing projects into specific sub-tasks before estimating is a partial intervention on scope neglect, which compounds the planning fallacy in complex projects. When the sub-tasks are individually estimated, the components that would have been invisible in a high-level estimate become visible and individually accountable. Review cycles, coordination overhead, handoffs, and rework all become estimable items rather than invisible assumptions.
Research suggests that detailed breakdown reduces underestimation by 25 to 30% compared to high-level estimates. It doesn't eliminate the planning fallacy, because the sub-task estimates are themselves subject to it, but it surfaces the hidden complexity that produces the largest overruns. The combination of task breakdown and reference class anchoring is the most accurate practical planning approach available without a formal forecasting system.
Planning buffers
Adding explicit contingency to estimates is a weaker intervention than RCF but a commonly accessible one. Adding 25 to 50% buffer to any estimate acknowledges the planning fallacy exists without precisely quantifying it. The weakness: buffers are subject to Parkinson's Law (work expands to fill the time available), so a buffered estimate often produces a project that takes as long as the buffered estimate rather than as long as the actual task requires.
Buffers work better as schedule design elements than as estimate corrections. Leaving 20 to 30% of a schedule unblocked, rather than adding 30% to each individual estimate, provides contingency without creating the Parkinson expansion risk at the individual task level. The buffer absorbs overruns structurally rather than inviting them by making every task duration larger.
Frequently asked questions
What is the planning fallacy?
The systematic tendency to underestimate how long tasks will take, how much they will cost, and how many things will go wrong. Identified by Kahneman and Tversky in 1979, it operates through the inside view: people estimate based on imagined optimistic scenarios rather than historical data from similar projects. It affects individuals, teams, corporations, and governments, and is not reliably corrected by experience or awareness.
Why doesn't experience fix the planning fallacy?
Experience does not fix the planning fallacy because overruns get attributed to specific external causes rather than incorporated as base rate evidence. The dependency that wasn't anticipated, the requirement that changed, the technical problem nobody predicted — each is explained as an anomaly. The next estimate begins fresh from an optimistic inside view without updating the underlying model. This attribution pattern is why expertise does not correct the bias.
What is reference class forecasting?
The most effective intervention on the planning fallacy. Instead of estimating how long this task will take (inside view), identify the category of similar tasks, establish the distribution of actual completion times for that category, and anchor the estimate to that distribution. Developed by Kahneman and formally applied to infrastructure planning, it consistently produces more accurate estimates than conventional planning because it forces the outside view rather than the optimistic inside view.
How do I apply reference class forecasting to my own work?
Track actual completion times for recurring task types over two to three weeks. Use that historical data as the basis for future estimates rather than imagined scenarios. If client proposals average six hours across five tracked instances, estimate six hours for the next one rather than the three hours the inside view scenario suggests. The personal correction factor derived from this data is more accurate than any general heuristic because it reflects your specific work and context.
Does adding buffer time to estimates fix the planning fallacy?
Partially, but with a risk. Buffers acknowledge the planning fallacy without precisely quantifying it. The weakness is Parkinson's Law: work expands to fill the time available, so buffered estimates often produce projects that take as long as the buffer rather than as long as the task requires. Leaving unscheduled slack in the overall schedule works better than adding buffer to each individual estimate, because it provides contingency without creating expansion incentives at the task level.
