The planning fallacy: why your two-week project takes five
You know your last three projects ran late. You are confident this one won't. That confidence has a name, and decades of research on how to beat it.
WrongThis time I've planned it carefully, so it'll take two weeks.
RightSimilar jobs have taken me four to six weeks, so I'll plan for five.
What people get wrong#
The planning fallacy is the tendency to underestimate how long your own tasks will take, even when you know that similar tasks have usually run long. The term was coined by Daniel Kahneman and Amos Tversky in 1979, as psychologist Roger Buehler recounts in a piece for the Society for Personality and Social Psychology.
The key word is own. People are not bad at estimating in general. They are specifically optimistic about their own plans while staying realistic, even cynical, about everyone else's.
The evidence#
The classic demonstration is a 1994 paper by Buehler, Griffin and Ross. They asked 37 psychology honors students when they expected to submit their theses. On average, the students' best estimate was 33.9 days. The actual average was 55.5 days. Fewer than a third (29.7%) finished by their own best-guess date.
It gets worse. The researchers also asked for a pessimistic estimate: when would you finish "if everything went as poorly as it possibly could"? The average worst case was 48.6 days, still a week short of reality, and fewer than half the students beat their own worst case.
The same paper opens with grander examples. The Sydney Opera House was estimated in 1957 to be finished early in 1963 for $7 million; a scaled-down version opened in 1973 at a cost of $102 million. Large projects are where the fallacy gets expensive, which is why planners study it closely: Bent Flyvbjerg's work on infrastructure forecasts documents how inaccurate cost and demand forecasts are, and attributes the errors to optimism bias and, sometimes, strategic misrepresentation.
Why it happens#
Buehler and colleagues tested a few explanations, and their results supported each one:
- We plan from the inside. When students thought aloud while forecasting, they focused mostly on future scenarios — how the work would go — and rarely on how past projects went.
- We explain away the past. Earlier misses were attributed to external, transient and specific causes, so they didn't seem relevant to the next prediction.
- Observers don't have the same blind spot. People predicting someone else's completion time were not optimistic. If anything, they overestimated, and they leaned more on that person's track record.
In other words, a plan is a story about things going right. Real projects contain interruptions, dependencies and surprises that no individual plan includes, but every past project did.
The fix#
"I've broken it into steps, so two weeks is realistic."
"My last three similar projects took four, five and six weeks. I'll plan for five."
Start from what comparable work actually took, then adjust.
Kahneman and Tversky's remedy, and the one Buehler highlights, is to take the outside view. In its formal version, reference class forecasting, you identify a set of similar past projects, look at how they actually turned out, and place your project within that distribution before touching your plan. Buehler's own Study 4 found that simply prompting people to connect relevant past experiences with their prediction eliminated the optimistic bias.
The SPSP piece lists two other families of tactics:
- Change how you imagine the plan. Break it into smaller steps (unpacking), plan backward from the deadline, or picture yourself doing it from an observer's point of view. Each forces obstacles into view.
- Make binding plans. Commit to doing specific parts at specific times and dates. These "implementation intentions" make you more likely to follow through.
A five-minute version#
Before you quote a date, write down the last three things you did that resemble this one and how long each really took. Use their average as your starting estimate, not your plan. If someone else is available, ask them to estimate how long it will take you. Their number is probably closer than yours.
Sources
Every factual claim on this page links to one of these. Checked October 23, 2026. See how we source.