Why tasks take longer than you think, and how to estimate better

Tasks take longer than you think because you picture the plan going smoothly and give little weight to how similar tasks went before. Psychologists call this the planning fallacy. In a classic 1994 study, students expected to finish their thesis in about 34 days and took about 56.

What is the planning fallacy?

The planning fallacy is predicting that a task will take less time than it does, even when you know similar tasks took longer. The best-known evidence comes from Roger Buehler, Dale Griffin and Michael Ross, who published five studies with university students in 1994, several of them relying on people reporting when they finished.¹

In the first, 37 psychology students in their final thesis semester at the University of Waterloo predicted when they would hand in their thesis. Among the 33 who had handed it in by the end of the records, the average prediction was 33.9 days. They took 55.5. Fewer than a third of all 37 finished by their own best estimate.¹

The researchers also asked for a worst case, the date “if everything went as poorly as it possibly could.” Even then, only 48.7% finished by that date.¹ Imagining the worst did not protect people from guessing short.

It happens with small tasks too. In a second study, students named tasks they meant to finish in the coming week. For the non-school ones, such as fixing a bicycle or cleaning an apartment, they expected 5.0 days and took 9.2.¹

One encouraging detail: the guesses still carried information. In the thesis study, people who predicted longer did take longer.¹ Our estimates are informative and consistently short.

Why do we guess short?

We build the estimate from the plan in our head, and we explain away the times things went over. In Buehler’s third study, students said their thoughts out loud while predicting when they would finish a school project. 74% of what they said was about the future, and 7% about relevant past experience.¹

Studies 3 and 4 in the same paper found a second habit: when people remembered running late before, they blamed outside, one-off causes, which made the past seem irrelevant to the next prediction.¹

A 2005 review by Michael Roy and colleagues adds another layer. It argues that people do base predictions on memories of similar tasks, and that those memories are themselves too short. “People appear to underestimate future event duration because they underestimate past event duration,” the authors write in the abstract.²

Does listing the steps help?

Listing the steps usually makes an estimate bigger, and for people who guess short that is a step toward the real figure. In three experiments from 2008, Forsyth and Burt asked people to allocate time either to a whole task or to each of its subtasks. The time given to the whole task was consistently smaller than the sum of the times given to its parts.⁴ They called this the segmentation effect.

The effect depends on what you list. In three studies from 2014, spelling out unusual parts of a task that take long raised estimates, spelling out unusual parts that are quick lowered them, and spelling out the typical parts made no difference.⁵

So the useful version is targeted. When you break a task down, look for the slow or easily forgotten parts: setting up, waiting, tidying away, the second pass.

What this can’t tell us. Whether a bigger estimate from listing steps was also a more accurate one in these tasks. That depends on how short your first guess was.

Why is your own track record the best estimate?

Because a record replaces the memory that runs short. In 2008, Roy, Mitten and Christenfeld asked college students to predict how long a task would take, such as counting sheets of paper or writing an essay. In two experiments, telling them how long a previous similar task had actually taken them increased the accuracy of their predictions.³ In a third, with people recruited in places like fast-food restaurants, giving them the average time other people had needed worked too.³

Buehler’s fourth study points the same way.¹ Students predicted when they would finish a one-hour computer assignment. One group was asked to recall how early they usually finished similar assignments and to describe how that would play out this time. That group’s average prediction matched their actual finishing time. Their predictions were less biased, though no more precise for each individual.

What does a better estimate look like?

Take a gut guess, list the parts, then check your record. Here is a worked example for “clean the kitchen”. The minutes are an illustration, not data.

How much buffer should you add?

Use your own ratio of actual to estimated time. Popular rules like “double it” or “add 20%” are easy to remember, and we found no study that tests those specific numbers.

Fill in the sheet for a week or two. For each task, divide the actual minutes by your guess. The middle value of those ratios is your personal buffer. If your typical ratio is 1.4, a 30-minute guess becomes 42 minutes in your calendar.

Keep separate ratios for kinds of work that differ a lot. A paper notebook, a spreadsheet or a timer you already own is enough. When you plan a whole day in blocks, the same ratio applies to each block; see time blocking for days that don’t go to plan.

Questions people ask

Sources

1. Buehler, R., Griffin, D., & Ross, M. (1994). Exploring the “planning fallacy”: Why people underestimate their task completion times. Journal of Personality and Social Psychology, 67(3), 366–381.FIVE STUDIES, INCLUDING EXPERIMENTS · UNIVERSITY STUDENTS · STUDIES 1–4: N = 37, 104, 78, 123doi.org/10.1037/0022-3514.67.3.366↩ a b c d e f g h i j
2. Roy, M. M., Christenfeld, N. J. S., & McKenzie, C. R. M. (2005). Underestimating the duration of future events: Memory incorrectly used or memory bias? Psychological Bulletin, 131(5), 738–756.NARRATIVE REVIEW · STUDIES OF DURATION ESTIMATESdoi.org/10.1037/0033-2909.131.5.738↩ a
3. Roy, M. M., Mitten, S. T., & Christenfeld, N. J. S. (2008). Correcting memory improves accuracy of predicted task duration. Journal of Experimental Psychology: Applied, 14(3), 266–275.THREE EXPERIMENTS · COLLEGE STUDENTS AND MEMBERS OF THE PUBLICdoi.org/10.1037/1076-898X.14.3.266↩ a b c
4. Forsyth, D. K., & Burt, C. D. B. (2008). Allocating time to future tasks: The effect of task segmentation on planning fallacy bias. Memory & Cognition, 36(4), 791–798.THREE EXPERIMENTS · TIME ALLOCATION TO WHOLE TASKS AND SUBTASKSdoi.org/10.3758/MC.36.4.791↩ a
5. Hadjichristidis, C., Summers, B., & Thomas, K. (2014). Unpacking estimates of task duration: The role of typicality and temporality. Journal of Experimental Social Psychology, 51, 45–50.THREE STUDIES · UNIVERSITY STUDENTS · STUDY 1 N = 152doi.org/10.1016/j.jesp.2013.10.009↩ a